<![CDATA[Hardware Acceleration in Robotics]]>https://news.accelerationrobotics.com/https://news.accelerationrobotics.com/favicon.pngHardware Acceleration in Roboticshttps://news.accelerationrobotics.com/Ghost 5.33Mon, 06 Apr 2026 18:30:46 GMT60<![CDATA[ROBOTCORE® Collaborative unveiled for Human-Robot Interaction Without Aggression. Hardware-Accelerated Dynamic Collision Avoidance Collaborative Behavior]]>https://news.accelerationrobotics.com/acceleration-robotics-announces-the-launch-of-robotcore-r-collaborative-revolutionizing-human-robot-interaction-with-hardware-accelerated-dynamic-collision-avoidance/66fe5745a4ced44a6aa3859aFri, 04 Oct 2024 05:05:53 GMTAcceleration Robotics Announces the Launch of ROBOTCORE® Collaborative to ensure Safe and Seamless Collaboration Between Humans and Robots with Unmatched Speed and Precision
Press release (English) link
Nota de prensa (Español) link

VITORIA-GASTEIZ, Spain – Acceleration Robotics, a leading robotics semiconductor startup based in the Basque Country, Spain, today announced the release of ROBOTCORE® Collaborative, a cutting-edge Human-Robot Interaction (HRI)-centric hardware solution that redefines the safety, speed, and efficiency of collaborative robotics. By incorporating advanced FPGA technology and a hardware-accelerated control system, ROBOTCORE® Collaborative ensures real-time dynamic collision avoidance, guaranteeing that robots and humans can work together seamlessly in shared workspaces without collisions or aggressive behaviors.

Dynamic collision avoidance: Advanced Human-Robot Interation without Aggression and with Accelerated Adaptive Evasion

Unparalleled Safety in Collaborative Robotics

0:00
/
ROBOTCORE® Collaborative is an innovative Human Robot Interaction (HRI)-centric hardware-accelerated control approach that ensures seamless and safe collaboration between robots and humans, without collisions or aggressive behaviors. By leveraging advanced Artificial Potential Field (APF) based control for safe and efficient robotic operation on industrial scenarios and FPGA technology, it provides unparalleled determinism and rapid response times, crucial for preventing collisions and aggressive behaviors in shared workspaces between humans and robots.

ROBOTCORE® Collaborative harnesses the power of Artificial Potential Field (APF)-based control and hardware acceleration to deliver microsecond-level response times, which are critical for preventing collisions and aggressive movements in industrial and dynamic environments. With this breakthrough, Acceleration Robotics offers an advanced platform that ensures deterministic behavior, even in unpredictable scenarios. By enabling real-time adjustments based on sensor data, the technology ensures a smooth and safe collaboration between robots and humans, transforming robotic arms into advanced collaborative systems.

"ROBOTCORE® Collaborative is a game-changer in the realm of human-robot interaction without aggression," said Victor Mayoral Vilches, founder of Acceleration Robotics. "Our hardware-accelerated approach offers the speed and determinism necessary to create truly collaborative environments where safety and performance are paramount."
0:00
/

Key Features of ROBOTCORE® Collaborative

  • Dynamic Collision Avoidance: The system leverages real-time data from depth cameras and sensors to detect obstacles and adjust the robot’s movements instantly, ensuring safe interaction.
  • Hardware-Accelerated Control: Powered by FPGA technology, ROBOTCORE® Collaborative provides rapid, real-time responses that outperform traditional CPU-based systems, ensuring microsecond-level latency.
  • ROS 2 Compatibility: Fully API-compatible with the ROS 2 control stack, this solution integrates seamlessly into existing robotic systems, making it easy for developers to enhance their robots with advanced collaborative behaviors.
  • Occlusion Handling: Multiple cameras extend the coverage of the robot's environment, allowing the system to handle occlusions and ensure continued safe operation even when some visual data is blocked.
0:00
/

Versatility Across Industries

ROBOTCORE® Collaborative is designed to meet the demands of a variety of industries, where safety, efficiency, and seamless human-robot interaction are critical:

  • Manufacturing: Enhances productivity and safety by enabling robots to work side by side with humans on assembly lines.
  • Healthcare: Facilitates safe collaboration between medical professionals and robots in surgical and caregiving environments.
  • Warehousing and Logistics: Ensures the safe coexistence of human and robotic workers in fast-paced, dynamic environments.
  • Research and Development: Provides a robust platform for advancing human-robot interaction technologies in academic and industrial research settings.

Unmatched Performance Powered by Hardware Acceleration

ROBOTCORE® Collaborative distance calculation operation | Min, Mean and Max Computation Time (ms) breakdown (logarithmic scale)

Compared to software-based solutions, ROBOTCORE® Collaborative delivers up to 29.49x faster processing times, reducing the average processing time to 37.7 milliseconds on FPGA, compared to over 1112 milliseconds on conventional CPUs. This level of performance is essential for real-time decision-making in collaborative robots (cobots), where precision and speed directly impact safety and efficiency.

Dynamic Collision Avoidance: The Future of Collaborative Robotics

ROBOTCORE® Collaborative introduces a new era of intelligent robotic systems that can navigate complex environments with ease. By leveraging depth image-based control and real-time data from multiple cameras, the system ensures accurate obstacle detection and avoidance, making it ideal for environments where human safety is critical.

For more information about ROBOTCORE® Collaborative, visit Acceleration Robotics or contact us to learn how this groundbreaking technology can enhance your industrial or research robots.

About Acceleration Robotics

Acceleration Robotics designs specialized semiconductors and hardware for robots to increase their performance and responsiveness. Leveraging GPUs and FPGAs, the company focuses on creating customized robotic systems that accelerate communication, control, and decision-making.

]]><![CDATA[Acceleration Robotics makes robots faster in La Rioja - Automotive Cluster Winners!]]>https://news.accelerationrobotics.com/acceleration-robotics-automotive-cluster-rioja-winners/6688020aa4ced44a6aa37f9eMon, 08 Jul 2024 10:08:22 GMT

ENG- We are happy to share that Acceleration Robotics won the first prize at the La Rioja Automotive Cluster awards as the #Autostartup 2024 the past 4 of July.

David Mayoral Vilches from Acceleration Robotics receiving the award for its deployment of hardware acceleration solutions in robotic setups. The company makes robots faster in industry.

We received excellent treatment from the entire organization, including Raül Llovet Camacho, Daniel Alfonso Alcántara, Miriam Saenz Perez, and the government of La Rioja with his president Gonzalo Capellan, its City Hall, Belinda León, presidents, and general directors.

On the left, we can see David Mayoral Vilches, sales officer at Acceleration Robotics, and on the right, we have Gonzalo Capellan, President of the University of La Rioja.
On the left, we can see Daniel Rueda Villadangos from the Automation Cluster; in the middle, David Mayoral Vilches, sales director at Acceleration Robotics; on the right, Raül Llovet Camacho from the Automation Cluster. and Miiriam Saenz Perez from the. automation cluster

With determination, strength, and innovation, we continue working to make La Rioja a center of excellence and reference in automation and robotization.4o

In summary, we highly recommend that any startup or company looking to engage with and do business in the automotive sector around the La Rioja region contact the cluster. Our experience couldn't have been better.

-
ESP- Estamos muy contentos de compartir que Acceleration Robotics ganó el primer premio en los premios del Clúster Automoción de La Rioja como la #Autostartup 2024 el pasado 4 de Julio.

Recibimos un trato excelente de toda la organización, incluyendo a Raül Llovet Camacho, Daniel Alfonso Alcántara, Miriam Saenz Perez y el gobierno de La Rioja con su presidente Gonzalo Capellán, su Alcaldía, Belinda León, presidentes y directores generales.

Con determinación, fuerza e innovación, seguimos trabajando para hacer de La Rioja un centro de excelencia y referencia en automatización y robotización.

En resumen, recomendamos encarecidamente a cualquier startup o empresa que busque involucrarse y hacer negocios en el sector automotriz en la región de La Rioja que se ponga en contacto con el clúster. Nuestra experiencia no podría haber sido mejor

]]><![CDATA[Hardware Accelerating ROS 2 Perception on the Qualcomm RB5 Robotics Kit]]>https://news.accelerationrobotics.com/ros2-qualcomm-hardware-acceleration/65e858e3a4ced44a6aa3768fThu, 07 Mar 2024 14:15:24 GMTThis article discusses breakthroughs with Qualcomm RB5 and ROS 2 Hardware Acceleration in Robotics Perception that are achieved by leveraging OpenCL to offload computation to the GPU. Benchmarking results hint speedup differences of up to 4x between ROS2 perception nodes.

In a previous article [1], we explored mapping ROS 2 perception pipeline’s computation graphs to FPGA-based hardware accelerators such as AMD FPGA SoC solutions with kernels running at 250MHz. This initial work has since then evolved [2] into a complete community, open contributions and  open standards that enable and provide vendor-agnostic hardware acceleration solutions around the ROS 2 community. Products have also been created as part of the endeavour, like Acceleration Robotics’ ROBOTCORE Framework, which allows building robot IP core accelerators (or robot cores) in a vendor-agnostic manner and while focusing on a ROS 2-centric development experience.

In this report, we further explore porting the open ROS 2 Hardware Acceleration Stack to GPU hardware accelerators from Qualcomm's premium-tier QRB5165 processor, designed to help build consumer, enterprise or industrial robots with on-device AI and 5G connectivity. In particular, we leverage the widely adopted Qualcomm® Robotics RB5 Development Kit.

QRB5165 processor is designed to help you build smarter and powerful consumer, enterprise or industrial robots with on-device AI and 5G connectivity and more.

The Qualcomm RB5 Robotics Kit is an innovative platform geared towards high-compute, low-power robotics and drone applications. It is equipped with the Qualcomm Kryo CPU, Adreno 650 GPU and the Hexagon DSP, enabling developers to combine high-performance heterogeneous compute to develop power-efficient as well as cost-effective robots.  

The Qualcomm® Robotics RB5 Development kit contains a robotics-focused development board and compliant with the 96Boards open hardware specification. Available comprehensive set of demo applications and tutorials to accelerate development of robotics applications.

We leverage the QRB5165’s Adreno 650 GPU to parallelize a subset of ROS 2’s image-processing pipeline using hardware acceleration, achieving speed-ups of upto 400%.

Figure 1: Visualisation of hardware acceleration results viz. ROS2 perception pipeline runtime on the Qualcomm Kryo 585 CPU and the OpenCL-accelerated implementation for the Qualcomm Adreno 650 GPU in the form of mean runtime in ms for (1) benchmark a1, (2) benchmark a1*, (3) benchmarks a2 and a5

Accelerating ROS2 Perception Pipelines:

Camera output often needs preprocessing to meet the input requirements of multiple different perception functions. These can include operations such as resizing and correcting for lens distortion. Image rectification and resizing are fundamental operations in computer vision pipelines, essential for tasks like stereo vision, object detection, and depth estimation. Traditionally, these operations have been performed using software-based algorithms, which can be computationally intensive and time-consuming, particularly for high-resolution images or video streams.

Figure 2 : Visualization of image rectification and image resizing pipeline in ROS2

By offloading these tasks to hardware-accelerated OpenCL kernels running on the Adreno 650 GPU, we were able to achieve significant performance gains with speed-ups of up to 400%.

Another benefit of offloading computation to the GPU is the ability to group image processing operations by passing data directly from one kernel to another. This mitigates the message transfer overhead during receiving input data along with the data transfer overhead from the CPU to GPU. The latter is tackled with the use of zero-copy data transfers, which enables us to map device (accelerator) memory to the host (CPU).

Our OpenCL implementation has two components:

  • The kernel running on the GPU,
  • The host code running on the CPU which controls the execution of the kernel through the ROS2 pipeline.

To accelerate image processing on the GPU, the images must be transferred to GPU memory. The traditional way of doing this is to copy over the images to the VRAM where the GPU can access it, perform parallelized operations on the GPU, and then copy over the result from the GPU to the CPU. However, this adds the overhead of transferring data between the GPU and the CPU over the PCIe bridge.

The OpenCL specification includes memory allocation flags and API functions that developers can use to create applications that can be used to eliminate extra copies during code execution. This is referred to as zero-copy behaviour. We were able to leverage zero-copy behaviour using Qualcomm-specific memory allocation flags in OpenCL. Furthermore, we hid the data transfer behind kernel execution through the use of multiple queues (PCIe lanes give us duplex memory transfer capabilities). This ensures that image processing happens asynchronously and that the data transfer operation overheads are hidden behind kernel runtime.

We were thus also able to significantly improve the performance of the perception pipeline (by more than 300%) by grouping image processing operations in a single node that transfers data between GPU kernels. We have enlisted this benchmark in (Table 1) as a1*: perception_2nodes_combined. The other benchmarks also enlisted below (defined in the Robotperf framework [4]) include a1: perception_2nodes, a2: image rectification and a5: image resizing. The benchmark that was run for this report was latency. (Figure 1) provides a visualisation of the results.

Table 1 : ROS2 perception pipeline runtime

Hardware Node Mean (speedup) RMS (speedup)
RB5 - Kryo 585 CPU a1 132.79 ms 132.95 ms
a2 20.5416 ms 21.2928 ms
a5 39.9251 41.4270
RB5- Adreno 650 GPU a1 66.5104 ms (Δ 2.0) 71.7921 ms (Δ 1.85)
a1* 38.9777 ms (Δ 3.4) 41.0983 ms (Δ 3.23)
a2 17.9097 ms (Δ 1.2) 19.8730 ms (Δ 1.1)
a5 28.7540 ms (Δ 1.4) 31.2265 ms (Δ 1.32)

References:

  1. Hardware accelerated ROS 2 pipelines and towards the Robotic Processing Unit (RPU)
  2. https://github.com/ros-acceleration/community/issues/39
  3. OpenCL Overview - The Khronos Group Inc
  4. https://github.com/robotperf/benchmarks/blob/main/benchmarks/README.md
]]><![CDATA[Hardware Acceleration in Robotics #91]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-91/6582f479a4ced44a6aa36fd5Fri, 22 Dec 2023 09:02:01 GMT

News

India’s first edition of ROSCon 2023 energises robotics community in country

Hardware Acceleration in Robotics #91

Source: The New Indian Express, December 19, 2023 [1]
The event featured seven innovative robotic startup showcases, 14 talks on ROS, and four-panel discussions.

Learnings from recent robotics investment data

Source: Robots & Startups , December 10, 2023 [2]
Robotics investment remains steady despite the challenging financial climate, with larger investors investing more in early-stage companies. However, the difficulty in raising from LPs is affecting younger and early-stage funds. In 2023, robotics investment is shifted towards fewer transactions and larger amounts, reflecting conservatism.

CHIPS act and U.S. fabs

Source: Semiwiki, December 19, 2023 [3]
The CHIPS Act, signed into law by President Biden in August 2022, provides $52.7 billion to the US semiconductor industry, including $39 billion in manufacturing incentives. The act, which began in November 2019, aims to build a wafer fab in the US, with the US government promising to provide subsidies.

Chinese chip companies are turning to Malaysia for their high-end chip, GPU needs

Source: Firstpost, December 18, 2023 [4]
China's semiconductor design and manufacturing companies are turning to Malaysian companies, for the assembly of high-end chips and GPUs. As the US tightens restrictions on the sale of high-end GPUs, smaller Chinese semiconductor firms are grappling with issues in advanced packaging

Source: Robotics & Automation, December 13, 2023 [5]
It is no longer a surprise that businesses are turning to robotics and automation to improve the efficiency of their daily operations – companies around the world have started to realise that these technologies are essential to remaining competitive.

Robot operating systems market to reach $1.8 billion by 2032 at 12.9% CAGR

Source: Yahoo Finance, December 14, 2023 [6]
Allied Market Research's report on the Robot Operating Systems Market reveals a global opportunity analysis and industry forecast for 2023-2032. The report predicts a CAGR of 12.9% from 2023 to 2032, with the industry generating $534.2 million in 2022 and $1.8 billion by 2032.

The use of semiconductors in automotive manufacturing

Source: AZoM, December 18, 2023 [7]
With the automotive sector aiming towards more reliable, efficient, and safe cars, the incorporation of semiconductors in vehicles has seen a major boost. The number of semiconductor chips in a modern car varies with the type; however, one can expect the number in the range of 1000 – 3500.

Hardware retailer adopts mobile robots for inventory control

Source: Mobile Robot Guide, December 8, 2023 [8]
Stine, a 13-store hardware retail chain in Louisiana and Mississippi, aims to improve operational efficiencies and customer service by leveraging robotics to offload manual tasks and ease labor shortages. Additionally, the retailer aims to improve on-shelf product availability through accurate data and analytics.

Papers

An optimized FPGA architecture for machine learning spplications

Source: Science Direct, December 4, 2023 [9]
This paper presents an FPGA architecture with posit multipliers that outperforms IEEE-754 multipliers in terms of delay and area. The proposed architecture outperforms the Stratix IV and Stratix 10 FPGAs in terms of critical path delay, with improvements of 25.5% and 43%, respectively, and consumes 10% less area and 22% less power.

REFRESH FPGAs: Sustainable FPGA Chiplet Architectures

Source: arXiv, November 28, 2023 [10]
The demand for agile computational power in edge and cloud infrastructure is increasing, but addressing the environmental impacts is crucial. A life-cycle view of sustainability is needed to reduce greenhouse gas emissions. FPGA architectures have not changed significantly, but REFRESH FPGAs can build new devices and architectures from retired dies using 2.5D integration. This approach leverages renewable energy integration into data centers, improving sustainability and amortizing cost investment over a longer lifetime.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. India’s first edition of ROSCon 2023 energises robotics community in country. (2023, December 19). The New Indian Express. https://www.newindianexpress.com/cities/bengaluru/2023/dec/19/indias-first-edition-of-roscon-2023energises-robotics-community-in-country-2642777.html ↩︎

  2. Keay, A. (2023, December 10). Learnings from recent robotics investment data. Robots & Startups . https://249x.substack.com/p/learnings-from-recent-robotics-investment ↩︎

  3. Jewell, B. (2023, December 19). CHIPS act and U.S. Fabs. Semiwiki. Retrieved December 20, 2023, from https://semiwiki.com/semiconductor-services/semiconductor-intelligence/339873-chips-act-and-u-s-fabs/ ↩︎

  4. Chinese chip companies are turning to Malaysia for their high-end chip, GPU needs. (2023, December 18). Firstpost. https://www.firstpost.com/tech/chinese-chip-companies-are-turning-to-malaysia-for-their-high-end-chip-gpu-needs-13517152.html ↩︎

  5. O'Sullivan-Dale, U. (2023, December 13). Robotics & automation’s 2024 trends forecast. Robotics & Automation. https://www.roboticsandautomationmagazine.co.uk/features/robotics-automations-2024-trends-forecast.html ↩︎

  6. Robot operating systems market to reach $1.8 billion by 2032 at 12.9% CAGR. (2023, December 14). Yahoo Finance. https://finance.yahoo.com/news/robot-operating-systems-market-reach-020000900.html ↩︎

  7. Abbasi, I. (2023, December 18). The use of semiconductors in automotive manufacturing. AZoM. https://www.azom.com/article.aspx?ArticleID=23253 ↩︎

  8. Oitzman, M. (2023, December 8). Hardware retailer adopts mobile robots for inventory control. Mobile Robot Guide. https://mobilerobotguide.com/2023/12/08/hardware-retailer-adopts-mobile-robots-inventory-control/?spMailingID=104669&puid=2502618&E=2502618&utm_source=newsletter&utm_medium=email&utm_campaign=104669 ↩︎

  9. An optimized FPGA architecture for machine learning spplications. (2023, December 4). Science Direct. https://www.sciencedirect.com/science/article/abs/pii/S1434841123004855 ↩︎

  10. Zhou, P., Zhuang, J., Cahoon, S., Tang, Y., Yang, Z., Chen, X., Shi, Y., Hu, J., & Jones, A. K. (2023, November 28). REFRESH FPGAs: Sustainable FPGA Chiplet Architectures. arXiv. https://arxiv.org/pdf/2312.02991.pdf ↩︎

]]><![CDATA[Hardware Acceleration in Robotics #90]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-90/6579cfe5a4ced44a6aa368adFri, 15 Dec 2023 08:01:14 GMT

News

Indian robotics is shackled by high duties. The government must free it

Hardware Acceleration in Robotics #90

Source: mint, December 6, 2023 [1]
India's quest to lead in robotics manufacturing faces a duty dilemma. While the nation dreams big with its National Strategy on Robotics, high import duties hinder the manufacturer’s ability to scale up production and export.

A venture capitalist's perspective on robotics

Source: The Robot Report, December 11, 2023 [2]
The robotics space changes rapidly, with constantly rotating “hot sectors,” tweaks to business models, and shifting investment and exit dynamics. To better understand the industry landscape, F-Prime Capital recently completed a comprehensive analysis of more than 1,250 robotics companies that raised funding over the past five years.

NVIDIA Isaac Sim on AWS to ease robotics development

Source: The Robot Report, December 12, 2023 [3]
NVIDIA has announced that its GPUs will be integrated into AWS to improve the Isaac simulator, offering a 2x performance boost. The new Amazon Machine Images (AMIs) on the NVIDIA L40S will enable roboticists to run Isaac Sim workloads, enabling real-time inferencing and fine-tuning of large language models.

Intel stands by record semiconductor investment despite German budget problems

Source: EURACTIV, December 5, 2023 [4]
Intel is sticking to its plans to build a €30 billion chip factory in Germany, despite some uncertainty over state subsidies for the project after the Constitutional Court ‘threw a spanner in the works’ of the government’s budget plans.

GPU market recovers - shipments up 37.4% in Q3 2023, with AMD seeing the biggest increase

Source: TweakTown, December 5, 2023 [5]
A new report shows that the GPU market is bouncing back in a big way, with shipments way up for AMD, Intel, and NVIDIA in Q3 2023.

EFPGA gives you FPGA speed and density at much less cost and power

Source: Semiconductor Engineering, December 7, 2023 [6]
FPGAs offer flexibility and quick market time, making them useful in various systems. As volumes grow, embedded FPGAs can be integrated to adapt to changing standards, improve algorithms, and optimize customer optimizations for SoC designers.

AI accelerator architectures poised for big changes

Source: Semiconductor Engineering, December 4, 2023 [7]
Design teams are racing to boost speed and energy efficiency of AI as it begins shifting toward the edge.

Unveiling the computing power: Apple’s M2 Max GPU against Nvidia’s titans

Source: Medium, December 12, 2023 [8]
Apple's M1 chip revolutionized high-end desktops with impressive performance and energy efficiency. The M2 Pro and M2 Max chips, particularly the M2 Max with 30 GPU cores, are expected to push boundaries further.

Papers

Vision-based robotics using open FPGAs

Source: Science Direct, November 14, 2023 [9]
Robotics offers practical applications in manufacturing, autonomous driving, and logistics. FPGAs are suitable computing substrates for robots due to their fast response time, low power consumption, parallelism, and flexibility. However, proprietary tools and hardware devices have fragmented the FPGA market. Open-source toolchains are emerging, offering a library of Verilog modules for vision-based robots. Three applications with real robots and open FPGAs have been developed for experimental validation. This opens up an attractive alternative to proprietary tools and FPGAs.

Multi-LiDAR localization and mapping pipeline for urban autonomous driving

Source: arXiv, November 3, 2023 [10]
A novel sensor fusion-based pipeline for offline and online localization of autonomous vehicles uses four LiDAR sensors and KISS-ICP mapping algorithms. The pipeline is integrated into the ROS 2 Autoware software stack, outperforming state-of-the-art approaches for real-world autonomous driving applications and research vehicles.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. Kaur, S., & Vikhe, G. (2023, December 6). Indian robotics is shackled by high duties. The government must free it. mint. https://www.livemint.com/opinion/first-person/government-must-slash-duties-and-taxes-to-make-indian-robotics-competitive-11701859794692.html ↩︎

  2. Aggarwal, S. (2023, December 11). A venture capitalist's perspective on robotics. The Robot Report. https://www.therobotreport.com/venture-capitalist-perspective-on-robotics/ ↩︎

  3. The Robot Report Staff. (2023, December 12). NVIDIA Isaac Sim on AWS to ease robotics development. The Robot Report. https://www.therobotreport.com/nvidia-isaac-sim-on-aws-to-ease-robotics-development/ ↩︎

  4. Alipour, N. (2023, December 5). Intel stands by record semiconductor investment despite German budget problems. EURACTIV. https://www.euractiv.com/section/politics/news/intel-stands-by-record-semiconductor-investment-despite-german-budget-problems/ ↩︎

  5. Andreadis, K. (2023, December 5). GPU market recovers - shipments up 37.4% in Q3 2023, with AMD seeing the biggest increase. TweakTown. https://www.tweaktown.com/news/94753/gpu-market-recovers-shipments-up-37-4-in-q3-2023-with-amd-seeing-the-biggest-increase/index.html ↩︎

  6. Tate, G. (2023, December 7). EFPGA gives you FPGA speed and density at much less cost and power. Semiconductor Engineering. https://semiengineering.com/efpga-gives-you-fpga-speed-and-density-at-much-less-cost-and-power/ ↩︎

  7. Mutschler, A. (2023, December 4). AI accelerator architectures poised for big changes. Semiconductor Engineering. https://semiengineering.com/ai-accelerator-architectures-poised-for-big-changes/?cmid=0f756a92-98dd-4c7e-b971-343c5339a553 ↩︎

  8. Harber, H. (2023, December 12). Unveiling the computing power: Apple’s M2 Max GPU against Nvidia’s titans. Medium. https://medium.com/@hyman.harber/unveiling-the-computing-power-apples-m2-max-gpu-against-nvidia-s-titans-ffb39e4408cb ↩︎

  9. Machado, F., Nieto, R., Fernández-Conde, J., Lobato, D., & Cañas, J. M. (2023, November 14). Vision-based robotics using open FPGAs. Science Direct. https://www.sciencedirect.com/science/article/pii/S0141933123002193 ↩︎

  10. Sauerbeck, F., Kulmer, D., Pielmeier, M., Leitenstern, M., Weib, C., & Betz, J. (2023, November 3). Multi-LiDAR localization and mapping pipeline for urban autonomous driving. arXiv. https://arxiv.org/pdf/2311.01823.pdf ↩︎

]]><![CDATA[Hardware Acceleration in Robotics #89]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-89/65705b01a4ced44a6aa362d5Fri, 08 Dec 2023 09:45:14 GMT

News

Acceleration robotics launches ROBOTCORE ROS 2 hardware

Hardware Acceleration in Robotics #89

Source: The Robot Report, December 1, 2023 [1]
One emerging protocol for real-time, deterministic communication in robotics is time-sensitive networking, or TSN. The key advantage of TSN is that it uses standard Ethernet networking cables, hubs, and switches.

Robotics investment for October 2023 totals $980 million

Source: Robotics Business Review, November 30, 2023 [2]
Robotics funding for the month of October 2023 equaled approximately $980 million, the result of 68 rounds. The October investments bring the 2023 robotics funding totals to $11.7 billion.

Robotics funding saw another dip in 2023

Source: TechCrunch, November 8, 2023 [3]
In 2021, robotics startups experienced a surge in interest due to supply chain issues and labor shortages. Despite being insulated from broader investment slowdowns, 2022 was the second-worst year for robotics investments in the past five years, second only to 2020, which was a global anomaly due to five consecutive quarters of decline in venture capital money.

Siemens and Intel to collaborate on advanced semiconductor manufacturing

Source: HPCwire, December 4, 2023 [4]
Siemens and Intel have signed a memorandum of understanding to collaborate on digitalization and sustainability of microelectronics manufacturing. The partnership will focus on advancing manufacturing efforts, evolving factory operations, cybersecurity, and supporting a resilient global industry ecosystem. Siemens will bring its IoT-enabled hardware, software, and electrical equipment.

Commission launches Chips Joint Undertaking under the European Chips Act

Source: European Commission, November 30, 2023 [5]
The Commission has inaugurated the Chips Joint Undertaking (Chips JU), aiming to strengthen the European semiconductor ecosystem and technological leadership. The initiative will bridge the gap between research, innovation, and production, facilitating commercialization of innovative ideas. The first call for funding is €1.67 billion, expected to be matched by Member States and private funds.

Warehouse automation market to return to growth in 2024

Source: The Robot Report, November 30, 2023 [6]
Interact Analysis predicts a resurgence in demand for warehouse automation in 2024, following a challenging 2023, with revenues rising and double-digit growth rates starting in 2025. The pandemic and low interest rates led to increased e-commerce orders and warehouse construction.

Japan partners with Nvidia to boost GPU stock pile

Source: Computerworld, December 5, 2023 [7]
Nvidia’s CEO is promising more support for Japan as all the nations in the world want more GPUs – and the US wants to dictate whom they are sold to.

Nvidia working closely with US to ensure new chips for China are compliant with curbs

Source: Reuters, December 6, 2023 [8]
Nvidia is collaborating with the US government to ensure new Chinese AI chips comply with export curbs, potentially allowing Chinese rivals to gain market share.

Papers

Design of the chassis and locomotion system for a mobile robot

Source: Victoria University of Wellington, October 10, 2023 [9]
MARVIN, the Mobile Autonomous Robotic Vehicle for Indoor Navigation, has undergone a transformation to integrate existing electronics for power supply, torso actuators, and sensor integration. An Intel NUC computer running ROS2 was integrated for precise control. The project also created dedicated ROS2 nodes for motor control, torso management, and sensor interaction. MARVIN is now a versatile platform for security and human-robot interaction research, showcasing sustainability and innovation in robotics technology.

Innovative solutions for fast autonomous navigation (SINAV)

Source: ResearchGate, October 19, 2023 [10]
This paper presents a continuous and autonomous navigation architecture developed by TAS-I and PoliTo for unstructured environments and uneven grounds, using active sensors and Deep Learning modules. The architecture also benefits from dark travel time. The focus is on algorithmic and computational aspects.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. Oitzman, M. (2023, December 1). Acceleration robotics launches ROBOTCORE ROS 2 hardware. The Robot Report. https://www.therobotreport.com/acceleration-robotics-launches-robotcore-ros-2-hardware/ ↩︎

  2. Kara, D. (2023, November 30). Robotics investment for October 2023 totals $980 million. Robotics Business Review. https://www.roboticsbusinessreview.com/rbr-investment/robotics-investment-for-october-2023-totals-980-million/ ↩︎

  3. Heater, B. (2023, November 8). Robotics funding saw another dip in 2023. TechCrunch. https://techcrunch.com/2023/11/08/robotics-funding-saw-another-dip-in-2023/ ↩︎

  4. Siemens and Intel to collaborate on advanced semiconductor manufacturing. (2023, December 4). HPCwire. https://www.hpcwire.com/off-the-wire/siemens-and-intel-to-collaborate-on-advanced-semiconductor-manufacturing/ ↩︎

  5. Commission launches Chips Joint Undertaking under the European Chips Act. (2023, November 30). European Commission. https://ec.europa.eu/commission/presscorner/detail/en/ip_23_6167 ↩︎

  6. Oitzman, M. (2023, November 30). Warehouse automation market to return to growth in 2024, says interact analysis. The Robot Report. https://www.therobotreport.com/warehouse-automation-market-will-grow-again-2024-says-interact-analysis/ ↩︎

  7. Reynolds, S. (2023, December 5). Japan partners with Nvidia to boost GPU stock pile. Computerworld. https://www.computerworld.com/article/3711502/japan-partners-with-nvidia-to-boost-gpu-stock-pile.html ↩︎

  8. Potkin, F. (2023, December 6). Nvidia working closely with US to ensure new chips for China are compliant with curbs. Reuters. https://www.reuters.com/technology/nvidia-develop-new-chips-that-comply-with-us-export-regulations-2023-12-06/ ↩︎

  9. Webber-LaHatte, A. (2023, October 10). Design of the chassis and locomotion system for a mobile robot. Victoria University of Wellington. https://ojs.victoria.ac.nz/wfes/article/view/8384 ↩︎

  10. Roncagliolo, P., Berardi, G., Lanza, P., Merlo, A., Graziato, D., & D’Amore, G. (2023, October 19). Innovative solutions for fast autonomous navigation (SINAV). ResearchGate. https://www.researchgate.net/profile/Patrick-Roncagliolo/publication/374813768_Innovative_Solutions_for_Fast_Autonomous_Navigation_SINAV/links/6530f9ac24bbe32d9a512173/Innovative-Solutions-for-Fast-Autonomous-Navigation-SINAV.pdf ↩︎

]]><![CDATA[Hardware Acceleration in Robotics #88]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-88/6567674ea4ced44a6aa35c2bFri, 01 Dec 2023 09:20:50 GMT

News

Semiconductors headed toward strong 2024

Hardware Acceleration in Robotics #88

Source: Semiwiki, November 20, 2023 [1]
The global semiconductor market is now solidly in a turnaround. WSTS revised its data for 2Q 2023 growth over 1Q 2023 to 6.0% from 4.2% previously. 3Q 2023 was up 6.3% from 2Q 2023. With our Semiconductor Intelligence forecast of 3% growth in 4Q 2023, the year-to-year growth in 4Q 2023 will be a positive 6%. This will set the stage for double-digit year-to-year growth for each quarter of 2024.

Intel will spend $25 billion on manufacturing its new chips at TSMC

Source: Tom's Hardware, November 29, 2023 [2]
Intel plans to place $14 billion worth of orders with TSMC in 2024 to fab 3nm CPU tiles, with orders totaling $10 billion in 2025. Intel is expected to become TSMC's second-largest 3nm customer, replacing AMD. By 2024, Intel expects to receive 15,000-3000 wafers per month from TSMC, increasing to 30,000 in 2025.

China’s pursuit of autonomous machine computing self-sufficiency

Source: The Diplomat, November 17, 2023 [3]
The development of computing chips for autonomous machines is China’s only missing link in completing its self-reliant autonomous machine ecosystem.

NVIDIA GPUs on AWS to offer 2x simulation LEAP in Omniverse Isaac Sim, accelerating smarter robots

Source: NVIDIA Blog, November 28, 2023 [4]
With NVIDIA L40S GPUs coming to Amazon Web Services, developers will be able to build and deploy robotics applications faster in the cloud using NVIDIA Isaac Sim.

Global data center GPU market analysis

Source: businesswire, November 21, 2023 [5]
The global data center GPU market is projected to grow from USD 14.3 billion in 2023 and is projected to reach USD 63 billion by 2028; it is expected to grow at a CAGR of 34.6% from 2023 to 2028.

Advanced applications of open-source technologies

Source: Express Computer, November 22, 2023 [6]
Open-source software has evolved from a hobbyist movement to a crucial component of cutting-edge technologies, offering accessible, customizable, and collaborative solutions. Its evolution has driven sophisticated applications across diverse domains, reshaping the digital landscape.

ROS the robot operating system in a nutshell

Source: Medium, November 26, 2023 [7]
Robot programming involves instructing a robot's brain, a computer, to perform tasks, connected to sensors and actuators. The computer, ranging from logic circuits to Raspberry Pi, plays a crucial role in the robot's operation.

Autonomous vehicles: Not ready yet

Source: Semiconductor Engineering, November 28, 2023 [8]
Much more R&D and real-world testing is needed to reach safe and secure L4/L5 technology.

Papers

Dynamic priority scheduling for periodic systems using ROS 2

Source: ECBS 2023, October 17, 2023 [9]
This paper proposes a dynamic priority scheduling algorithm for ROS 2 systems, based on callback deadlines and channel update rates, demonstrating its efficacy in reducing buffer size compared to conventional single-threaded executors.

Model-based generation of hardware/software architectures with hybrid schedulers for robotics systems

Source: IEEE Xplore, 2023 [10]
This work proposes a modeling approach for robotic systems that uses FPGA-based architectures to process compute-intensive algorithms. The approach aims to create compact, expressive descriptions of the system, integrating existing algorithms and leveraging FPGA advantages. The model must be application-independent and leverage FPGA advantages, surpassing software-based solutions. Data type and flow analysis are performed, and six schedulers are proposed for various scenarios. Solutions are evaluated using four use cases.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. Jewell, B. (2023, November 20). Semiconductors headed toward strong 2024. Semiwiki. https://semiwiki.com/semiconductor-services/338264-semiconductors-headed-toward-strong-2024/ ↩︎

  2. Connatser, M. (2023, November 29). Intel will spend $25 billion on manufacturing its new chips at TSMC: Report. Tom's Hardware. https://www.tomshardware.com/news/intel-will-spend-14-billion-on-manufacturing-chips-at-tsmc-report ↩︎

  3. Shaoshan, L. (2023, November 17). China’s pursuit of autonomous machine computing self-sufficiency. The Diplomat. https://thediplomat.com/2023/11/chinas-pursuit-of-autonomous-machine-computing-self-sufficiency/ ↩︎

  4. Andrews, G. (2023, November 28). NVIDIA GPUs on AWS to offer 2x simulation LEAP in Omniverse Isaac Sim, accelerating smarter robots. NVIDIA Blog. https://blogs.nvidia.com/blog/gpu-aws-omniverse-isaac-sim-robots/ ↩︎

  5. Global data center GPU market analysis. (2023, November 21). businesswire. https://www.businesswire.com/news/home/20231121212113/en/Global-Data-Center-GPU-Market-Analysis-Report-2023-2028-with-Case-Studies-from-NVIDIA-Imagination-Technologies-and-Mobica-Realtek-Intel-and-Google---ResearchAndMarkets.com ↩︎

  6. Advanced applications of open-source technologies. (2023, November 22). Express Computer. https://www.expresscomputer.in/guest-blogs/advanced-applications-of-open-source-technologies-darshil-shah/105983/ ↩︎

  7. Karthik, K. (2023, November 26). ROS the robot operating system in a nutshell. Medium. https://medium.com/@karthikbyju2486/ros-the-robot-operating-system-in-a-nutshell-a89dd3922172 ↩︎

  8. Koon, J. (2023, November 28). Autonomous vehicles: Not ready yet. Semiconductor Engineering. https://semiengineering.com/autonomous-vehicles-not-ready-yet/?cmid=4677c769-f4be-4543-922a-9eee6c3f1ce6 ↩︎

  9. Dust, L., & Mubeen, S. (2023, October 17). Dynamic priority scheduling for periodic systems using ROS 2. ECBS 2023. https://www.idt.mdu.se/ecbs2023/pre-prints/paper_4150.pdf ↩︎

  10. Podlubne, A., Mey, J., Andreou, A., Pertuz, S., Aßmann, U., & Göhringer, D. (2023). Model-based generation of hardware/software architectures with hybrid schedulers for robotics systems. IEEE Xplore. https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10284570 ↩︎

]]><![CDATA[Hardware Acceleration in Robotics #87]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-87/655e223aa4ced44a6aa35604Fri, 24 Nov 2023 08:55:50 GMT

News

Acceleration Robotics announces ROBOTCORE® ROS 2 and RTPS, boosting ROS 2 communications from 62x to thousands-fold faster

Hardware Acceleration in Robotics #87

Source: Acceleration Robotics, November 21, 2023 [1]
Unleashing Unprecedented Speed in Robotics, Acceleration Robotics' ROBOTCORE® ROS 2 and RTPS Pave the Way Towards Robot-specific Chips Setting New Benchmarks in Performance, Energy-efficiency and Reliability for ROS 2 Networking.

Acceleration Robotics unveils ROBOTCORE® UDP/IP for ultra-fast, deterministic robot networking communications

Source: Acceleration Robotics, November 20, 2023 [2]
ROBOTCORE® UDP/IP is a Trailblazing FPGA Robot IP Core Resolving Networking Indeterminism in Robotics and Elevating Robotic Networking to New Heights of Speed.

The Program Agenda for ROSCon India is out

Source: , November 1, 2023 [3]
ROSCon India is the official domestic ROSCon for India, held by the local ROS community with the support of Open Robotics. Targeted at the local Indian robotics community, it provides an opportunity for ROS developers, hobbyists, researchers and industry professionals alike, to spend two days together learning from and networking with each other.

Microsoft adds FPGA-powered network accelerator to Azure

Source: The Register, November 21, 2023 [4]
Microsoft has introduced Azure Boost, a cloud hardware upgrade that aims to enhance the performance of all future instance types in its big blue cloud. Azure Boost offloads server virtualization processes onto purpose-built software and hardware, freeing up CPU resources for guest virtual machines.

Boosting custom ROS graphs using NVIDIA Isaac transport for ROS

Source: NVIDIA Technical Blog, November 20, 2023 [5]
NVIDIA Isaac Transport for ROS (NITROS) utilizes two hardware-acceleration features: type adaptation and type negotiation. Type adaptation optimizes data format for specific hardware accelerators, while type negotiation allows ROS nodes to advertise supported types for optimal performance.

Open robotics releases gazebo harmonic simulator

Source: The Robot Report, November 21, 2023 [6]
Open Robotics recently released Gazebo Harmonic, which is the eighth version of its open-source simulator. The company said this is one of its biggest Gazebo releases to date.

AMD delivers high performance, high efficiency CPUs and GPUs for all HPC use cases

Source: HPCwire, November 15, 2023 [7]
High-performance computing demands complex real-world problems and faster simulations, leading to increased demand for AMD solutions. AMD powers the fastest supercomputer and eight of the top ten most energy-efficient supercomputers on the Green500 list. The Frontier supercomputer at Oak Ridge National Laboratory is the fastest, passing the exaflop barrier with only 22.70 MW of energy consumption, making it #8 on the list.

China's hoard of chip-making tools: National treasures or expensive spare parts?

Source: Semiwiki, November 20, 2023 [8]
In the film Raiders of the Lost Ark, a warehouse filled with artifacts is compared to China's stockpiled semiconductor production equipment. China has exceeded production levels, with sales reaching 46% and 48% of total revenues in the third quarter, according to ASML and Lam Research.

Papers

Source: Aalborg University's Research Portal, October 2, 2023 [9]
The paper introduces a real-time human interaction detection system using RGB-D cameras for context-aware navigation in mobile robots. The system uses a convolutional neural network architecture and localizes interactions using human detection. It integrates detected human interaction zones into robot navigation costmaps, allowing for social spaces-aware planning.

Design and implementation of low-power, high-speed, reliable and secured Hardware Accelerator using 28 nm technology for biomedical devices

Source: Science Direct, October 10, 2023 [10]
This paper presents a Hardware Accelerator (HA) for Wireless Body Sensor Networks (WBSNs) using 28 nm technology on Zynq system-on-a-chip (SOC). The HA is implemented using Vivado and VITIS high-level synthesis tool, integrating with Xilinx tools. The HA improves 81% in speed compared to existing methods and enhances power and area of the WBSN, demonstrating the potential of 28 nm technology in wireless sensor networks.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. https://news.accelerationrobotics.com/robotcore-ros-2-and-rtps-ultrafast-networking-communications/ ↩︎

  2. https://news.accelerationrobotics.com/robotcore-udp-ip-fast-robot-networking-communications/ ↩︎

  3. https://rosconindia.in ↩︎

  4. Sharwood, S. (2023, November 21). Microsoft adds FPGA-powered network accelerator to Azure. The Register. https://www.theregister.com/2023/11/21/azure_boost_network_accelerator/?utm_source=daily&utm_medium=newsletter&utm_content=article ↩︎

  5. Bhide, A. (2023, November 20). Boosting custom ROS graphs using NVIDIA Isaac transport for ROS. NVIDIA Technical Blog. https://developer.nvidia.com/blog/boosting-custom-ros-graphs-using-nvidia-isaac-transport-for-ros/ ↩︎

  6. The Robot Report Staff. (2023, November 21). Open robotics releases gazebo harmonic simulator. The Robot Report. https://www.therobotreport.com/open-robotics-gazebo-harmonic-simulator/ ↩︎

  7. Ware, A. (2023, November 15). AMD delivers high performance, high efficiency CPUs and GPUs for all HPC use cases. HPCwire. https://www.hpcwire.com/2023/11/20/amd-delivers-high-performance-high-efficiency-cpus-and-gpus-for-all-hpc-use-cases/ ↩︎

  8. Addison, C. (2023, November 20). China's hoard of chip-making tools: National treasures or expensive spare parts? Semiwiki. Retrieved November 22, 2023, from https://semiwiki.com/lithography/338218-chinas-hoard-of-chip-making-tools-national-treasures-or-expensive-spare-parts/ ↩︎

  9. Schwörer, T., Schmidt, J. E., & Chrysostomou, D. (2023, October 2). Nav2CAN: Achieving context aware navigation in ROS2 using Nav2 and RGB-D sensing. Aalborg University's Research Portal. https://vbn.aau.dk/en/publications/nav2can-achieving-context-aware-navigation-in-ros2-using-nav2-and ↩︎

  10. Mavinkattimath, S., Khanai, R., Torse, D., & Iyer, N. (2023, October 10). Design and implementation of low-power, high-speed, reliable and secured Hardware Accelerator using 28 nm technology for biomedical devices. Science Direct. https://www.sciencedirect.com/science/article/abs/pii/S1746809423009874 ↩︎

]]><![CDATA[Acceleration Robotics announces ROBOTCORE® ROS 2 and RTPS, Boosting ROS 2 Communications from 62x to Thousands-Fold Faster]]>https://news.accelerationrobotics.com/robotcore-ros-2-and-rtps-ultrafast-networking-communications/655c869aa4ced44a6aa350c8Tue, 21 Nov 2023 14:21:32 GMTUnleashing Unprecedented Speed in Robotics, Acceleration Robotics' ROBOTCORE® ROS 2 and RTPS Pave the Way Towards Robot-specific Chips Setting New Benchmarks in Performance, Energy-efficiency and Reliability for ROS 2 Networking
Press release (English) link
Nota de prensa (Español) link

VITORIA-GASTEIZ, Spain – Acceleration Robotics —a robotics semiconductor startup founded in the Basque Country, Spain and with offices in India— announced the results after one year of development with the release of ROBOTCORE® ROS 2 and ROBOTCORE® RTPS, marking a significant milestone in the field of robotics networking by implementing in hardware (prototyped with an FPGA) technology that allows robots to exchange information in less than 2.5 microseconds. These products are designed to revolutionize robotics and ROS 2 networking communications, delivering speeds from 62x to thousands of times faster than current standards.

Round-Trip Average Network Latency (us) breakdown. ROBOTCORE® ROS 2 significantly enhances the networking architecture of robotic systems and tackles one the widely critizied issues of ROS 2: its latency overhead over the DDS communication middleware. It does so by building a hardware implementation of the core ROS 2 abstraction layers (RCL, RMW) and by establishing direct hardware datapaths with the underlying DDS middleware implementation (also possible in hardware, through  ROBOTCORE® RTPS). This allows to remove the latency overhead of ROS 2 over DDS and to deliver unmatched speed and absolute determinism.

ROBOTCORE® ROS 2: A New Era of High-Speed Robotics Networking

From 62x to Thousands-Fold Faster ROS 2. ROBOTCORE® ROS 2 operates by offloading its operations through advanced FPGA hardware acceleration technology, providing unmatched determinism and latency guarantees. When combined with ROBOTCORE UDP/IP and ROBOTCORE RTPS, it implements a hardware IP and UDP stack, as well as hardware DDS-RTPS processing mechanism, ensuring microsecond-level (us) rapid handling of network packages. This optimization results in significantly faster robot data transmission or reception over multiple data link layers (both wired and wireless), leading to reduced latency, crucial for real-time applications in robotics. ROBOTCORE

ROBOTCORE® ROS 2 is an expertly crafted hardware design that implements the ROS 2 robotics framework. Similarly, ROBOTCORE RTPS implements in hardware the Real-Time Publish Subscribe (RTPS) protocol and parts of the Data Distribution Service (DDS) communication middleware, a relevant communication backbone of robotics. Leveraging the power of FPGA technology for quick prototyping and implementation, this robot core (also known as IP core), is engineered to dramatically enhance the speed of ROS 2 robot networking communications. This means faster data processing, reduced latency, and improved synchronization across your robotics systems. Powering the future of accelerated robotics networking, ROBOTCORE® ROS 2 can send or receive packages in less than 2.5 microseconds, accelerating networking by more than 62x on average when compared to traditional software implementations on modern CPUs.

ROS 2 is instrumental in robotics and among its capabilities, it delivers a message-passing infrastructure for robots that acts like a communication 'databus', a system that connects and transfers data between robotic components over the network (unlike a database that stores data). ROBOTCORE® ROS 2 delivers faster robot data transmission or reception over multiple data link layers (both wired and wireless), leading to reduced latency and more determinism, crucial for real-time applications in robotics. ROBOTCORE® ROS 2 can send or receive small ROS 2 messages in less than 2.5 us (one-way) and establish a round-trip communication within 5 us for control applications, accelerating networking by more than 62x when compared to traditional software implementations on modern CPUs. When considering the worst case scenario (maximum latency) for small ROS 2 messages, ROBOTCORE ROS 2 remains deterministic, reporting a consistent round-trip latency below 5 us. Opposed to this, software implementations on modern CPUs take on average 339 us for the same task, but can reach latencies of 22.7 ms (22700 us) or more, which means ROBOTCORE ROS 2 is Thousands-Fold (4540x) faster in these worst case scenarios.

ROBOTCORE® ROS 2 delivers absolute determinism via hardware when combined with ROBOTCORE® RTPS and ROBOTCORE® UDP/IP. Compared to software-based solutions, it ensures that the communication latency is lower and that remains always the same, regardless of the load of the system. Crucial for real-time robotic systems and solving major bottlenecks in robotic communications[1],[2],[3],[4].

5 years pursuing better robot hardware

ROBOTCORE® ROS 2 is the result of the last 5 years working towards faster and more efficient hardware for robots. In 2018, Mayoral-Vilches and his team released a series of papers[1][2][3][4] that investigated robot networking and unveiled that modern CPU-based networking solutions, even when optimized, could not cope with many of the real-time needs of modern distributed robotic systems in industry. Custom computer architectures meant for robotics were in need. For the last years Mayoral-Vilches has been working with top experts from industry and academia – including Intel, AMD, Microchip or Harvard among others– to create tools to develop such systems and performance benchmarking. First with ROBOTCORE[5] and later with RobotPerf[6], this work has attracted hundreds across industry and academia due to its relevance for the future of robotics computing.

ROBOTCORE® ROS 2 is the latest result from this series and delivers absolute determinism via hardware when combined with underlying ROBOTCORE® RTPS and ROBOTCORE® UDP/IP designs. Compared to software-based solutions, it ensures that the communication latency is lower, consuming less energy and that latency remains always the same, regardless of the load of the system. Crucial for real-time robotic systems and solving major bottlenecks in robotic communications today.

  1. Gutiérrez, C. S. V., Juan, L. U. S., Ugarte, I. Z., & Vilches, V. M. (2018). Real-time Linux communications: an evaluation of the Linux communication stack for real-time robotic applications. arXiv preprint arXiv:1808.10821. ↩︎

  2. Gutiérrez, C. S. V., Juan, L. U. S., Ugarte, I. Z., & Vilches, V. M. (2018). Towards a distributed and real-time framework for robots: Evaluation of ROS 2.0 communications for real-time robotic applications. arXiv preprint arXiv:1809.02595. ↩︎

  3. Gutiérrez, C. S. V., Juan, L. U. S., Ugarte, I. Z., & Vilches, V. M. (2018). Time-sensitive networking for robotics. arXiv preprint arXiv:1804.07643. ↩︎

  4. Gutiérrez, C. S. V., Juan, L. U. S., Ugarte, I. Z., Goenaga, I. M., Kirschgens, L. A., & Vilches, V. M. (2018). Time synchronization in modular collaborative robots. arXiv preprint arXiv:1809.07295. ↩︎

  5. Mayoral-Vilches, V., Neuman, S. M., Plancher, B., & Reddi, V. J. (2022, October). ROBOTCORE: An open architecture for hardware acceleration in ROS 2. In 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (pp. 9692-9699). IEEE. ↩︎

  6. Mayoral-Vilches, V., Jabbour, J., Hsiao, Y. S., Wan, Z., Martínez-Fariña, A., Crespo-Alvarez, M., ... & Reddi, V. J. (2023). RobotPerf: An Open-Source, Vendor-Agnostic, Benchmarking Suite for Evaluating Robotics Computing System Performance. arXiv preprint arXiv:2309.09212 ↩︎

Key Features and Technological Advancements

  • Microsecond-Level Speed: Achieves unprecedented speeds in networking communications, vital for real-time robotic applications.
  • Significant Latency Reduction: Compared to traditional CPUs, ROBOTCORE® ROS 2, RTPS and UDP/IP drastically lowers latency, ensuring rapid and efficient data transmission.
  • Deterministic Networking: Addresses the lack of determinism in common robotic networking stacks, providing consistent and predictable communication performance.
  • Energy and Resource Efficiency: Balances high-speed data processing with low power consumption, essential for energy-conscious environments.
Sending a ROS 2 message with ROBOTCORE ROS 2 takes about 1.775 microjoules (uJ). ROBOTCORE® ROS 2 stands out remarkably, outperforming contemporary workstation CPU and GPU solutions when it comes to power performance. ROBOTCORE® ROS 2 exhibits an exceptional efficiency, being 500 times more power-efficient compared to modern workstation CPUs and GPUs. For comparison, a single neuron of a human brain firing one action potential takes roughly 0.03 to 0.3 microjoules (uJ), while the energy required to Blink an Eye is roughly estimated to be around 1 millijoule (mJ). These similarities underscore that ROBOTCORE® ROS 2 operates with an energy budget comparable to the fundamental biological processes in the human brain.

Versatility in Applications

The ROBOTCORE® ROS 2 is not just technologically superior; it is also versatile and widely compatible. Ideal for scenarios where speed and reliability are non-negotiable, it finds extensive applications in:

  • Industrial Automation: Streamlines communication in manufacturing and assembly lines for enhanced operational efficiency.
  • Remote Operation: Offers smooth and responsive control in teleoperation systems, crucial for precision tasks.
  • Autonomous Vehicles: Ensures rapid data exchange essential for the real-time decision-making of autonomous driving systems.
  • Research and Development: Provides a reliable platform for developing and testing next-generation robotic technologies.

Supported Silicon and Hardware Solutions: Broad Compatibility for Diverse Applications

ROBOTCORE® ROS 2 is not only a breakthrough in technology but also in versatility. Designed to operate seamlessly with the most popular FPGA silicon vendors, it offers broad compatibility with FPGA and FPGA SoC solutions from leading silicon vendors including Intel, AMD or Microchip, supporting new silicon solutions like Intel's Agilex, AMD’s Versal or Microchip’s PolarFire. This also includes a range of development kits, empowering developers to build advanced robots with hardware acceleration and ROS (Robot Operating System).

About Acceleration Robotics

Acceleration Robotics is a firm focused on designing customized brains for robots to hasten their response time. Founded by top robotic experts to deliver semiconductor building blocks for robots, the company leverages GPUs and FPGAs to create custom hardware that speeds up a robot's operation.

]]><![CDATA[Acceleration Robotics unveils ROBOTCORE® UDP/IP for Ultra-Fast, Deterministic Robot Networking Communications]]>https://news.accelerationrobotics.com/robotcore-udp-ip-fast-robot-networking-communications/655b1a34a4ced44a6aa34c3dMon, 20 Nov 2023 08:54:12 GMTROBOTCORE® UDP/IP is a Trailblazing FPGA Robot IP Core Resolving Networking Indeterminism in Robotics and Elevating Robotic Networking to New Heights of Speed
Press release (English) link
Nota de prensa (Español) link

VITORIA-GASTEIZ, Spain – Acceleration Robotics —a robotics semiconductor startup founded in the Basque Country, Spain and with offices in India— today unveils the ROBOTCORE® UDP/IP. Building on the success of its innovative ROBOTCORE® series, this cutting-edge FPGA robot core (IP core) is specifically designed to revolutionize robot networking communications through hardware acceleration. Leveraging the User Datagram Protocol (UDP) for Internet Protocol (IP) stack communications, this product sets a new benchmark for speed and reliability in robotic systems.

Unprecedented Speed and Reliability

ROBOTCORE® UDP/IP brings a paradigm shift in networking communications within the robotics industry. With the ability to send or receive small packages in under 1 microsecond (<700 nanoseconds for 8-byte payloads) and 1024-byte packets in just 3 microseconds, it outpaces traditional CPU-based solutions dramatically. This performance translates to an average latency speedup of more than 19x when compared to traditional CPUs, and in cases of maximum latency for real-time compliance, the speedup exceeds 80x. Visit benchmarks for more information on performance.

Prior research[1],[2],[3],[4] helped uncover that traditional networking stacks in robotics have long struggled with indeterminism. ROBOTCORE® UDP/IP delivers absolute determinism via hardware. When compared to software-based solutions, it ensures that the communication latency is always the same, regardless of the load of the system. Crucial for real-time robotic systems.

Key Features and Technological Advancements

  • Microsecond-Level Speed: Achieves unprecedented speeds in networking communications, vital for real-time robotic applications.
  • Significant Latency Reduction: Compared to traditional CPUs, ROBOTCORE® UDP/IP drastically lowers latency, ensuring rapid and efficient data transmission.
  • Deterministic Networking: Addresses the lack of determinism in common robotic networking stacks, providing consistent and predictable communication performance.
  • Energy and Resource Efficiency: Balances high-speed data processing with low power consumption, essential for energy-conscious environments.


Versatility in Applications

The ROBOTCORE® UDP/IP is not just technologically superior; it is also versatile and widely compatible. Ideal for scenarios where speed and reliability are non-negotiable, it finds extensive applications in:

  • Industrial Automation: Streamlines communication in manufacturing and assembly lines for enhanced operational efficiency.
  • Remote Operation: Offers smooth and responsive control in teleoperation systems, crucial for precision tasks.
  • Autonomous Vehicles: Ensures rapid data exchange essential for the real-time decision-making of autonomous driving systems.
  • Research and Development: Provides a reliable platform for developing and testing next-generation robotic technologies.

Supported Silicon and Hardware Solutions: Broad Compatibility for Diverse Applications

ROBOTCORE® UDP/IP is not only a breakthrough in technology but also in versatility. Designed to operate seamlessly with the most popular FPGA silicon vendors, it offers broad compatibility with FPGA and FPGA SoC solutions from leading silicon vendors including Intel, AMD or Microchip, supporting new silicon solutions like Intel's Agilex, AMD’s Versal or Microchip’s PolarFire. This also includes a range of development kits, empowering developers to build advanced robots with hardware acceleration and ROS (Robot Operating System).

Optimized for ROS: Elevating Networking in Robotics with ROBOTCORE® UDP/IP

ROBOTCORE® UDP/IP significantly boosts ROS and ROS 2 systems, offering microsecond-level speed for enhanced real-time responsiveness. Seamlessly integrating with the common ROS workflow, it provides accelerated networking capabilities while maintaining familiar development processes.

ROBOTCORE® UDP/IP operates by accelerating the communication speed of the Internet Protocol (IP) networking stack through advanced FPGA hardware acceleration specifically tailored for UDP (User Datagram Protocol) communications. At its core, it implements a streamlined and optimized hardware IP and UDP stack processing mechanism, ensuring microsecond-level (us) rapid handling of network packets. This optimization results in significantly faster robot data transmission and reduced latency, crucial for real-time applications in robotics.

A Commitment to Faster Robots

"Our prior research helped uncover that traditional networking stacks in robotics have long struggled with indeterminism, posing significant challenges in high-precision and real-time applications. With ROBOTCORE® UDP/IP, we've specifically targeted this issue, delivering a solution that not only enhances the speed of robot communications, but also ensures predictable and consistent communication performance. This is more than an innovation; it's a fundamental shift in how we approach robotic networking interactions, opening new possibilities in the field,"
stated Víctor Mayoral-Vilches, CTO and founder of Acceleration Robotics.

About Acceleration Robotics

Acceleration Robotics is a firm focused on designing customized brains for robots to hasten their response time. Founded by top robotic experts to deliver semiconductor building blocks for robots, the company leverages GPUs and FPGAs to create custom hardware that speeds up a robot's operation.

]]><![CDATA[Hardware Acceleration in Robotics #86]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-86/6554ddbca4ced44a6aa3462fFri, 17 Nov 2023 09:24:36 GMT

News

ROSCon 2023 talks

Hardware Acceleration in Robotics #86

Source: ROSCon, November 9, 2023 [1]
The videos of the ROSCon 2023 talks are now available. ROSCon 2023 is a chance for ROS developers of all levels, beginner to expert, to spend an extraordinary two days learning from and networking with the ROS community.

The European Innovation Council has approved €1 billion in deep tech venture investments in one year

Source: European Commission, November 14, 2023 [2]
The Commission announced that nearly €1 billion of investments in deep tech companies have been approved by the European Innovation Council (EIC) Fund since the Fund started its operations in September 2022.

California is the robotics capital of the world

Source: Robots & Startups , November 11, 2023 [3]
China has fewer robotics startups than the USA. The UK has more than Germany, France or Switzerland and other surprising facts about the real robotics revolution.

Agile robots acquires Franka Emika

Source: The Robot Report, November 14, 2023 [4]
German robotics developer Agile Robots is acquiring Franka Emika GmbH, two months after filing for preliminary insolvency. Agile will take over operations, keeping its 100-person staff, and plans to grow the company, which filed for insolvency due to shareholder differences.

An FPGA-to-ASIC case study for refining smart meter design

Source: EDN, November 9, 2023 [5]
FPGAs are often used in embedded system designs for prototyping or software development. However, they are often converted to ASICs for volume manufacturing. This process can be nearly automatic, but may be challenging if multiple chips are integrated or mixed-signal functions are needed.

Are robot chips the key to humanoid robots' development?

Source: CGTN, November 10, 2023 [6]
How important are chips to robots? How do they make robots smart, especially humanoid robots? At the sixth China International Import Expo, CGTN interviewed Shi Ying, the field application engineer director at Texas Instruments.

Intrinsic demonstrates interoperability framework integration with KEBA robot controller

Source: The Robot Report, November 14, 2023 [7]
Intrinsic Innovation LLC is building a software platform that it said will make robotics more accessible to developers. This week, the company is demonstrating how its real-time control framework can be integrated with KEBA Group AG’s robot controllers at the Smart Production Solutions, or SPS, trade fair.

Chinese AI and robotics companies spearhead digital transformation in the middle east

Source: China Money Network, November 11, 2023 [8]
China's AI companies, SenseTime and Terminus, are expanding globally, particularly in the Middle East. SenseTime, a Chinese AI company, has collaborated with local authorities to introduce inclusive AI educational courses and collaborated with the Saudi Arabian Public Investment Fund to establish SenseTime MEA. Terminus, a Chinese AIoT service provider, built the 2020 Dubai Expo venue as a smart city using IoT logic.

Papers

Source: Aalborg University's Research Portal, 2023 [9]
The paper introduces a real-time human interaction detection system for mobile robots using RGB-D cameras. The system uses a convolutional neural network architecture and localizes human interactions in 3D scenes. It integrates these zones into robot navigation costmaps, allowing socially-aware navigation in human environments. The system has been validated through simulated and real-world tests.

Design and implementation of low-power, high-speed, reliable and secured hardware accelerator using 28 nm technology for biomedical devices

Source: ScienceDirect, October 10, 2023 [10]
This paper presents a Hardware Accelerator (HA) for Wireless Body Sensor Networks (WBSNs) using 28 nm technology on Zynq system-on-a-chip (SOC). The HA is implemented using Vivado and VITIS high-level synthesis tool, integrating with Xilinx tools. The HA improves 81% in speed compared to existing methods and enhances power and area of the WBSN, demonstrating the potential of 28 nm technology in wireless sensor networks.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. https://vimeo.com/showcase/roscon2023 ↩︎

  2. The European Innovation Council has approved €1 billion in deep tech venture investments in one year. (2023, November 14). European Commission. https://ec.europa.eu/commission/presscorner/detail/en/ip_23_5728 ↩︎

  3. Keay, A. (2023, November 11). California is the robotics capital of the world. Robots & Startups . https://249x.substack.com/p/california-is-the-robotics-capital ↩︎

  4. Wessling, B. (2023, November 14). Agile robots acquires Franka Emika. The Robot Report. https://www.therobotreport.com/agile-robots-acquires-franka-emika/ ↩︎

  5. Lai, B. (2023, October 9). An FPGA-to-ASIC case study for refining smart meter design. EDN. https://www.edn.com/an-fpga-to-asic-case-study-for-refining-smart-meter-design/ ↩︎

  6. CGTN. (2023, November 10). Are robot chips the key to humanoid robots' development? CGTN . https://news.cgtn.com/news/2023-11-10/Are-robot-chips-the-key-to-humanoid-robots-development--1oBQfHHYXkc/index.html ↩︎

  7. The Robot Report Staff. (2023, November 14). Intrinsic demonstrates interoperability framework integration with KEBA robot controller. The Robot Report. https://www.therobotreport.com/intrinsic-demonstrates-integration-with-keba-robot-controller-at-smart-production-solutions/ ↩︎

  8. Chinese AI and robotics companies spearhead digital transformation in the middle east. (2023, November 11). China Money Network. https://www.chinamoneynetwork.com/2023/11/11/chinese-ai-and-robotics-companies-spearhead-digital-transformation-in-the-middle-eastChina's AI companies, SenseTime and Terminus, are expanding globally, particularly in the Middle East. SenseTime, a Chinese AI company, has collaborated with local authorities to introduce inclusive AI educational courses and collaborated with the Saudi Arabian Public Investment Fund to establish SenseTime MEA. Terminus, a Chinese AIoT service provider, built the 2020 Dubai Expo venue as a smart city using IoT logic. ↩︎

  9. Schwörer, T., Schmidt, J. E., & Chrysostomou, D. (in press). Nav2CAN: Achieving Context Aware Navigation in
    ROS2 Using Nav2 and RGB-D sensing. In IEEE International Conference on Imaging Systems and Techniques
    (IST 2023) IEEE ↩︎

  10. Mavinkattimath, S., Khanai, R., Torse, D., & Iver, N. (2023, October 10). Design and implementation of low-power, high-speed, reliable and secured hardware accelerator using 28 nm technology for biomedical devices. ScienceDirect. https://www.sciencedirect.com/science/article/abs/pii/S1746809423009874 ↩︎

]]><![CDATA[Hardware Acceleration in Robotics #85]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-85/654d2adba4ced44a6aa33daaFri, 10 Nov 2023 08:06:18 GMT

News

China plans to mass produce humanoids by 2025

Hardware Acceleration in Robotics #85

Source: The Robot Report, November 7, 2023 [1]
China's Ministry of Industry and Information Technology (MIIT) has set ambitious goals for developing humanoid robots, with a goal to mass-produce them by 2025. The plan includes establishing a humanoid innovation system, making key technology breakthroughs, and ensuring core component supply.

Vietnam sets sights on becoming semiconductor hub

Source: EE Times, November 7, 2023 [2]
Vietnam's ambition to become a key semiconductor hub in the region has been fueled by U.S. President Biden's visit and strategic partnership announcement. FPT Information System (FPT IS) has established a subsidiary, FPT Semiconductor, to focus on power management ICs (PMICs) design and development.

Nvidia plans to release three new chips for China

Source: Reuters, November 9, 2023 [3]
Nvidia plans to release three new chips for China, following the US's ban on selling two AI chips and one gaming chip to Chinese firms. The chips are HGX H20, L20 PCIe, and L2 PCIe.

EFPGA architectural improvements that lower test cost and increase quality

Source: Semiconductor Engineering, November 2, 2023 [4]
Pipelining key portions of the scan circuitry to increase scan speed.

CUDA-accelerated robot motion generation in milliseconds with NVIDIA cuRobo

Source: NVIDIA Technical Blog, November 7, 2023 [5]
Fast motion-generation algorithms for real-time autonomous robot navigation can be used in various industries like food and services, warehouse automation, and machine tending. However, this requires complex constraints and minimizing costs, requiring local optimization.

AI used in construction of chip, microchip facilities

Source: EE Times, November 8, 2023 [6]
The semiconductor and microchip manufacturing industry is rapidly evolving, with the integration of artificial intelligence transforming construction methods and transforming the development of high-tech manufacturing facilities.

How arm is gaining chip dominance with its architecture in Apple, Nvidia, AMD, Amazon, Qualcomm and more

Source: CNBC, November 9, 2023 [7]
Arm, the dominant company in chip architecture, powers nearly every smartphone, with Apple relying on custom silicon for iPhones and MacBooks. Its IPO in September valued it above $54 billion due to increasing companies opting for Arm over Intel's x86 architecture.

Global semiconductor sales increase 1.9% month-to-month in september

Source: SIA, November 1, 2023 [8]
The Semiconductor Industry Association (SIA) reported a 1.9% increase in global semiconductor sales in September 2023, with a total of $134.7 billion in Q3. This represents a 6.3% increase from Q2 2023 and a 4.5% decrease from Q3 2022. The SIA represents 99% of the U.S. semiconductor industry and nearly two-thirds of non-U.S. chip firms.

Papers

Photonic accelerators for image segmentation in autonomous driving and defect detection

Source: arXiv, October 3, 2023 [9]
Photonic computing offers faster and more energy-efficient deep neural network inference, impacting applications like autonomous driving and defect detection. This paper investigates image segmentation on photonic accelerators, focusing on the best DNN architectures and their throughput and energy efficiency. It demonstrates that certain segmentation models exhibit negligible accuracy loss when executed on photonic accelerators, and discusses techniques for recovering accuracy in poor models.

Survey of simulators for aerial robots

Source: arXiv, November 4, 2023 [10]
This paper analyzes existing UAV simulators and decision factors to enhance efficiency and safety in research endeavors. It highlights the challenges of selecting the best simulator for specific use-cases, despite the numerous available options.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. Wessling, B. (2023, November 7). China plans to mass produce humanoids by 2025. The Robot Report. https://www.therobotreport.com/china-plans-to-mass-produce-humanoids-by-2025/?spMailingID=102502&puid=2502618&E=2502618&utm_source=newsletter&utm_medium=email&utm_campaign=102502 ↩︎

  2. Dahad, N. (2023, November 7). Vietnam sets sights on becoming semiconductor hub. EE Times. https://www.eetimes.com/vietnam-sets-sights-on-becoming-semiconductor-hub/ ↩︎

  3. Nvidia plans to release three new chips for China - local media. (2023, November 9). Reuters. https://www.reuters.com/technology/nvidia-plans-release-three-new-chips-china-local-media-2023-11-09/ ↩︎

  4. Tate, G. (2023, November 2). EFPGA architectural improvements that lower test cost and increase quality. Semiconductor Engineering. https://semiengineering.com/efpga-architectural-improvements-that-lower-test-cost-and-increase-quality/?cmid=eca05a35-c19a-4026-913e-d83df1d8700a ↩︎

  5. Sundaralingam, B., & Hari, S. (2023, November 7). CUDA-accelerated robot motion generation in milliseconds with NVIDIA cuRobo. NVIDIA Technical Blog. https://developer.nvidia.com/blog/cuda-accelerated-robot-motion-generation-in-milliseconds-with-curobo/ ↩︎

  6. Morkos, R. (2023, November 8). AI used in construction of chip, microchip facilities. EE Times. https://www.eetimes.com/embracing-ai-in-construction-of-chip-microchip-facilities/ ↩︎

  7. Tarasov, K. (2023, November 9). How arm is gaining chip dominance with its architecture in Apple, Nvidia, AMD, Amazon, Qualcomm and more. CNBC. https://www.cnbc.com/2023/11/09/how-arm-gained-chip-dominance-with-apple-nvidia-amazon-and-qualcomm.html ↩︎

  8. Global semiconductor sales increase 1.9% month-to-month in september. (2023, November 1). SIA. https://www.semiconductors.org/global-semiconductor-sales-increase-1-9-month-to-month-in-september/ ↩︎

  9. Nair, L., Widemann, D., Turcott, B., Moore, N., Wleklinski, A., Bunandar, D., Papavasileiou, I., Wang, S., & Logan, E. (2023, October 3). Photonic accelerators for image segmentation in autonomous driving and defect detection. arXiv. https://arxiv.org/pdf/2309.16783.pdf ↩︎

  10. Dimmig, C. A., Silano, G., McGuire, K., Gabellieri, C., Honig, W., Moore, J., & Kobilarov, M. (2023, November 4). Survey of simulators for aerial robots. arXiv. https://arxiv.org/abs/2311.02296 ↩︎

]]><![CDATA[Hardware Acceleration in Robotics #84]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-84/6542945ca4ced44a6aa336f9Fri, 03 Nov 2023 08:27:07 GMT

News

September 2023 robotics investments total $1.84 billion

Hardware Acceleration in Robotics #84

Source: Robotics Business Review, October 31, 2023 [1]
September 2023 robotics funding totaled $1.84 billion, the result of 50 investments. Total funding for the months of January 2023 through September 2023 equals approximately $10.7 billion.

Thoughts on ROSCon 2023

Source: Weekly Robotics Newsletter, October 29, 2023 [2]
This post summarizes some of the highlights of the event.

Reflections on ROS in Asia

Source: LinkedIn, October 16, 2023 [3]
The International Federation of Robotics reported that a whopping 73% of “all newly deployed robots” last year were installed in Asia.

AMD sees profit surge as it banks on AI chips

Source: Silicon Republic, November 1, 2023 [4]
The company predicts that its MI300 AI accelerator chip will be the fastest product to reach $1bn in sales ‘in AMD history’.

System-on-Chip integration complexity and hardware/Software contracts

Source: Semiconductor Engineering, October 26, 2023 [5]
Control and status register mismanagement can lead to expensive oversights.

Bringing tiny Chiplets to embedded socs

Source: EE Times, October 27, 2023 [6]
ZeroASIC has developed a technology platform to bring chiplets to embedded system design, as a time-efficient alternative to designing and manufacturing custom application-specific ICs (ASICs).

What will that chip cost?

Source: Semiconductor Engineering, November 1, 2023 [7]
Establishing the true cost to develop an advanced chip is complicated, but headline numbers appear to be significantly inflated.

Decoding the functions of a processor

Source: The New Indian Express, November 1, 2023 [8]
These tiny chips are responsible for executing instructions, performing complex calculations, and enabling seamless functions on computers and smartphones.

Papers

Enabling HW-based task scheduling in large Multicore architectures

Source: IEEE Xplore, October 12, 2023 [9]
Dynamic Task Scheduling is a programming model that simplifies parallel program development, reducing scheduling overheads. Traditional SW-only systems have limited capacity for core count and task granularity. This study presents an FPGA-proven, Linux-capable 30-core RISC-V system with hardware accelerated Task Scheduling, demonstrating its competitive performance at high core counts. The architecture includes hardware and software optimizations, making it more scalable than previous solutions.

GateSeeder: Near-memory CPU-FPGA acceleration of short and long read mapping

Source: arXiv, September 29, 2023 [10]
Read mapping is a costly and time-consuming process in genomics analyses, primarily affecting Index Querying, Seed Chaining, and Sequence Alignment. This work aims to improve these steps by utilizing new algorithms and hardware devices. GateSeeder, the first CPU-FPGA-based near-memory acceleration for both short and long read mapping, uses modern FPGAs' near-memory computation capability and a lightweight algorithm for finding potential matching segment pairs. Experimental results show GateSeeder outperforms Minimap2 without sequence alignment.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. Kara, D. (2023, October 31). September 2023 robotics investments total $1.84 billion. Robotics Business Review. https://www.roboticsbusinessreview.com/rbr-investment/september-2023-robotics-investments-total-1-84-billion/ ↩︎

  2. Sadowski, M. (2023, October 29). Thoughts on ROSCon 2023. Weekly Robotics Newsletter . https://www.weeklyrobotics.com/articles/2023_10_29_roscon_2023/ ↩︎

  3. Festo, M. (2023, October 16). Reflections on ROS in Asia. LinkedIn. https://www.linkedin.com/pulse/reflections-ros-asia-matthew-festo-jvmgc/ ↩︎

  4. Gowran, L. (2023, November 1). AMD sees profit surge as it banks on AI chips. Silicon Republic. https://www.siliconrepublic.com/business/amd-earnings-profit-ai-accelerator-chips ↩︎

  5. Schirrmeister, F. (2023, October 26). System-on-Chip integration complexity and hardware/Software contracts. Semiconductor Engineering. https://semiengineering.com/system-on-chip-integration-complexity-and-hardware-software-contracts/ ↩︎

  6. Ward-Foxton, S. (2023, October 27). Bringing tiny Chiplets to embedded socs. EE Times. https://www.eetimes.com/bringing-tiny-chiplets-to-embedded-socs/ ↩︎

  7. Bailey, B. (2023, November 1). What will that chip cost? Semiconductor Engineering. https://semiengineering.com/what-will-that-chip-cost/ ↩︎

  8. Raj, A. (2023, November 1). Decoding the functions of a processor. The New Indian Express. https://www.newindianexpress.com/xplore/2023/nov/01/decoding-the-functions-of-a-processor-2628957.html ↩︎

  9. Morais, L., Álvarez, C., Jiménez-González, D., Miguel de Haro, J., Araujo, G., Frank, M., Goldman, A., & Martorell, X. (2023, October 12). Enabling HW-based task scheduling in large Multicore architectures. IEEE Xplore. https://ieeexplore.ieee.org/abstract/document/10284544 ↩︎

  10. Eudine, J., Alser, M., Singh, G., Alkan, C., & Mutlu, O. (2023, September 29). GateSeeder: Near-memory CPU-FPGA acceleration of short and long read mapping. arXiv. https://arxiv.org/pdf/2309.17063.pdf ↩︎

]]><![CDATA[Hardware Acceleration in Robotics #83]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-83/65392985a4ced44a6aa33022Fri, 27 Oct 2023 06:58:27 GMT

News

India’s path to becoming a global semiconductor hub: A transformative journey

Hardware Acceleration in Robotics #83

Source: Manufacturing Today India, October 23, 2023 [1]
India is strategically positioning itself as a global semiconductor hub, driven by government initiatives, high-level industry engagements, and emphasis on transformative technologies like FPGAs, despite the challenges in supply chain management and ecosystem development.

ROS Robotics Companies

Source: Github, October 23, 2023 [2]
A public list of companies that are known to use the Robot Operating System (ROS and ROS 2) or any of its related tools for development, to create products, to offer services or who ship ROS with or as part of their product(s). The list now has 650+ companies.

The path to Chiplet architecture

Source: Semiwiki, October 19, 2023 [3]
The semiconductor industry is thriving in advanced packaging systems, which combine multiple dies in a single package. Initially, 3D packaging had no effective design tools, but successful product working designs have been achieved.

Building openness into industrial robotics

Source: Intrinsic, October 19, 2023 [4]
Gathering with the developer community at ROSCon 2023. At ROSCon, the Open Robotics team at Intrinsic introduced the progress they’re making on the development of alternative middleware protocols as well as their first integrations between ROS and the Intrinsic platform.

Accelerate AI-enabled robotics with advanced simulation and perception tools on NVIDIA Isaac platform

Source: NVIDIA Technical Blog, October 19, 2023 [5]
NVIDIA has released major updates to its NVIDIA Isaac Robotics platform, including NVIDIA Isaac ROS 2.0 and NVIDIA Isaac Sim 2023.1, which provide high-fidelity simulation and perception capabilities for robotics developers. These updates aim to simplify the building and testing of AI-based robotic applications.

Nvidia expands robotics platform to ‘meet the rise of generative AI’

Source: Robotics & Automation News, October 23, 2023 [6]
Generative AI is bringing the power of transformer models and large language models to virtually every industry. That reach now includes areas that touch edge, robotics and logistics systems: defect detection, real-time asset tracking, autonomous planning and navigation, human-robot interactions and more.

Intel announces intent to operate PSG as standalone business

Source: Intel, October 3, 2023 [7]
Sandra Rivera to lead PSG as CEO. Decision positions PSG to more effectively compete in the FPGA market; ongoing strategic alignment with Intel drives value creation for both companies.

Source: Semiwiki, October 24, 2023 [8]
Electronic device production is recovering after a post-pandemic slump in late 2021. Smartphone shipments declined from -6% in 3Q 2021 to -18% in 4Q 2022. Canalys estimates 3Q 2023 shipments were down 1%, with a mid-single-digit year-to-year increase.

Papers

Accelerated object detection on FPGA SoC

Source: Marmara, August, 2023 [9]
The thesis aims to improve object detection speed, accuracy, and energy efficiency using two- and three-dimensional data on Field Programmable Gate Array (FPGA). Microprocessor-based systems have lower performance due to lack of parallel data processing capacity. Machine learning structures with many layers, neurons, are used for object detection but require high memory.

GPU acceleration of high-precision homomorphic computation utilizing redundant representation

Source: Cryptology ePrint Archive, September, 2023 [10]
Fully homomorphic encryption (FHE) allows secure data analysis, but its performance is lower for high-precision computations. This paper extends Klemsa and Önen's fast homomorphic algorithms for high-precision integers to multithreaded environments, resulting in faster performance. The approach performs 128-bit addition, 32-bit multiplication, and 128-bit Max and ReLU functions in 1.4 seconds using a Tesla V100S server.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. Kumar, S. (2023, October 23). India’s path to becoming a global semiconductor hub: A transformative journey. Manufacturing Today India. https://www.manufacturingtodayindia.com/sectors/indias-path-to-becoming-a-global-semiconductor-hub-a-transformative-journey ↩︎

  2. https://github.com/vmayoral/ros-robotics-companies ↩︎

  3. McLellan, P. (2023, October 19). The path to Chiplet architecture. Semiwiki. https://semiwiki.com/eda/siemens-eda/336451-the-path-to-chiplet-architecture/ ↩︎

  4. McCarter, J. (2023, October 19). Building openness into industrial robotics. Intrinsic. https://intrinsic.ai/blog/posts/building-openness-into-industrial-robotics/ ↩︎

  5. Andrews, G. (2023, October 19). Accelerate AI-enabled robotics with advanced simulation and perception tools on NVIDIA Isaac platform. NVIDIA Technical Blog. https://developer.nvidia.com/blog/accelerate-ai-enabled-robotics-with-advanced-simulation-and-perception-tools-in-nvidia-isaac-platform/ ↩︎

  6. Edwards, D. (2023, October 23). Nvidia expands robotics platform to ‘meet the rise of generative AI’. Robotics & Automation News. https://roboticsandautomationnews.com/2023/10/23/nvidia-expands-robotics-platform-to-meet-the-rise-of-generative-ai/73258/ ↩︎

  7. Intel announces intent to operate PSG as standalone business. (2023, October 3). Intel. https://www.intel.com/content/www/us/en/newsroom/news/intel-oct-2023-news.html#gs.6u03h0 ↩︎

  8. Jewell, B. (2023, October 24). Electronics production trending up. Semiwiki. https://semiwiki.com/semiconductor-services/337124-electronics-production-trending-up/ ↩︎

  9. Kara, M. A. (2023, August). Accelerated object detection on FPGA SoC. Marmara. https://openaccess.marmara.edu.tr/entities/publication/d3964176-1fdd-4171-ae09-d71a0a1101f4 ↩︎

  10. Narisada, S., Okada, H., Fukushima, K., Kiyomoto, S., & Nishide, T. (2023, September). GPU acceleration of high-precision homomorphic computation utilizing redundant representation. Cryptology ePrint Archive. https://eprint.iacr.org/2023/1467.pdf ↩︎

]]><![CDATA[Hardware Acceleration in Robotics #82]]>https://news.accelerationrobotics.com/hardware-acceleration-in-robotics-82/652feaa2a4ced44a6aa328feFri, 20 Oct 2023 09:08:18 GMT

News

Working with robots can make humans put in less effort

Hardware Acceleration in Robotics #82

Source: New Scientist, October 18, 2023 [1]
Robots that do their job reliably can cause humans who work alongside them to be less diligent because of a phenomenon called social loafing.

Google and Qualcomm team up for wearable devices chip

Source: Silicon Republic, October 18, 2023 [2]
The new chip powered by open-source RISC-V technology could help level the playing field in the contentious semiconductor space.

Why AMD’s upcoming chips won’t be the savior AI startups are hoping for

Source: The Information, October 11, 2023 [3]
Sharon Zhou, cofounder of Lamini, revealed using over 100 chips from Advanced Micro Devices for the past year, sparking excitement among AI founders and investors. This news suggests AMD could capitalize on the demand for AI chips, potentially taking market share from Nvidia.

Tormach to unveil cutting-edge real-time motion control in ROS at ROSCon 2023

Source: Robotics & Automation News, October 12, 2023 [4]
Tormach, a US-based CNC machine and industrial robot manufacturer, will make its inaugural appearance at ROSCon 2023, taking place in New Orleans from October 18 to 20.

Startups in the robotics industry

Source: Robotics & Automation News, October 13, 2023 [5]
Every robotics startup starts with just a clever idea. A clever idea that will bring technology to a specific industry, use case or application and answer the growing need for automation, due in particular to worldwide human aging and the labor shortage it creates, but also for re- and nearshoring demand based on global supply-chain issues.

Chip industry talent shortage drives academic partnerships

Source: Semiconductor Engineering, October 16, 2023 [6]
Universities, companies, and governments are forming broad partnerships to update skills and foster innovation in chips, security, AI, and related fields.

US to enforce new restrictions on China chips

Source: E+T Magazine, October 318 2023 [7]
The Biden administration has announced further measures preventing the sale of advanced semiconductor technology to China, including Nvidia’s AI chips.

RISC-V wants all your cores

Source: Semiconductor Engineering, October 12, 2023 [8]
It is not enough to want to dominate the world of CPUs. RISC-V has every core in its sights, and it’s starting to take steps to get there.

Papers

Toward composing efficient FPGA-based hardware accelerators for physics-based model predictive control smart sensor for HEV battery cell management

Source: IEEE Xplore, October 3, 2023 [9]
The depletion of fossil resources and climate change necessitate the development of sustainable transportation systems, such as hybrid electric vehicles (HEVs). Battery technology is crucial for HEVs, and model predictive control (MPC) and physics-based model (PBM) can help extend battery life. However, the computational complexity of physics-based MPC makes it infeasible for portable BMS on resource-constrained embedded devices. This research introduces a novel FPGA-based embedded hardware accelerator for PB-MPC smart sensors, achieving a 58 times speedup compared to its software counterpart and maintaining a small footprint for portable systems.

MICP-L: Mesh-based ICP for robot localization using hardware-accelerated ray casting

Source: ResearchGate, September 26, 2023 [10]
Mesh ICP Localization (MICP-L) is a computationally efficient method for robots to accurately localize themselves in 6D environments, even in GPS-denied ones. It uses range sensors to register them to a triangle mesh map, accelerating computation of ray casting correspondences. The approach is applicable in real-time on various target architectures and is demonstrated with datasets from agriculture, drones, and automotive domains.

Previous Hardware Acceleration in Robotics Newsletters

Past ROS 2 Hardware Acceleration Working Group meetings

  1. Sparkes, M. (2023, October 18). Working with robots can make humans put in less effort. New Scientist. https://www.newscientist.com/article/2397628-working-with-robots-can-make-humans-put-in-less-effort/ ↩︎

  2. Darmody, J. (2023, October 18). Google and Qualcomm team up for wearable devices chip. Silicon Republic. https://www.siliconrepublic.com/machines/google-qualcomm-chip-wearable-devices ↩︎

  3. Palazzolo, S. (2023, October 11). Why AMD’s upcoming chips won’t be the savior AI startups are hoping for. The Information. https://www.theinformation.com/articles/why-amds-upcoming-chips-wont-be-the-savior-ai-startups-are-hoping-for ↩︎

  4. Allinson, M. (2023, October 12). Tormach to unveil cutting-edge real-time motion control in ROS at ROSCon 2023. Robotics & Automation News. https://roboticsandautomationnews.com/2023/10/12/tormach-to-unveil-cutting-edge-real-time-motion-control-in-ros-at-roscon-2023/72945/ ↩︎

  5. Edwards, D. (2023, October 13). Startups in the robotics industry. Robotics & Automation News. https://roboticsandautomationnews.com/2023/10/13/startups-in-the-robotics-industry/73025/ ↩︎

  6. Allan, L. (2023, October 16). Chip industry talent shortage drives academic partnerships. Semiconductor Engineering. https://semiengineering.com/chip-industry-talent-shortage-drives-academic-partnerships/?cmid=d63df6e1-afcd-440e-ad0b-51efc3f51031 ↩︎

  7. Valero de Urquia, B. (2023, October 18). US to enforce new restrictions on China chips. E+T Magazine. https://eandt.theiet.org/content/articles/2023/10/us-sets-new-restrictions-on-china-chips/ ↩︎

  8. Bailey, B. (2023, October 12). RISC-V wants all your cores. Semiconductor Engineering. https://semiengineering.com/risc-v-wants-all-your-cores/?cmid=83a0609a-8f5d-4772-abb4-65dd1a5c5c25 ↩︎

  9. Madsen, A. K., & Perera, D. G. (2023, October 3). Toward composing efficient FPGA-based hardware accelerators for physics-based model predictive control smart sensor for HEV battery cell management. IEEE Xplore. https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=10262302 ↩︎

  10. Mock, A., Wiemann, T., Putz, S., & Hertzberg, J. (2023, September 26). MICP-L: Mesh-based ICP for robot localization using hardware-accelerated ray casting. ResearchGate. https://www.researchgate.net/profile/Alexander-Mock-5/publication/364732894_MICP-L_Mesh-based_ICP_for_Robot_Localization_using_Hardware-Accelerated_Ray_Casting/links/6512cc2437d0df2448edc899/MICP-L-Mesh-based-ICP-for-Robot-Localization-using-Hardware-Accelerated-Ray-Casting.pdf ↩︎

]]>