RoboARCH: Robotics Acceleration with Computing Hardware and Systems

Autonomous robots must act in real-time within dynamic environments, requiring robust performance across diverse scales. To do so, roboticists are increasingly leveraging heterogeneous deployments across edge and cloud infrastructures and integrating and co-designing robotics algorithms for specific computing hardware ranging from CPUs, to GPUs, to FPGAs, to MCUs, and even to custom accelerator ASICs (e.g., Neuromorphic processors). As such, a key challenge for the field of robotics in the future will be to determine not only the right allocation of algorithms to computer hardware and the right combination of heterogeneous computer hardware onboard robotic systems to balance performance with power, but also the right software engineering practices and paradigms to broaden access to these advanced approaches across the robotics community. As such, this workshop is designed to build a welcoming community and ignite innovation and collaboration at the exciting intersection of robotics and computer software, systems, and architecture, focusing on the development of advanced robotic systems through cutting-edge computational solutions for the benefit of all. Featuring talks and panels from distinguished speakers in academia and industry, the workshop will explore challenges and opportunities presented by emerging computing technologies in high-performance robotics. The workshop will also showcase posters and lightning talks from students and professionals, highlighting recent advances and providing a platform for sharing ideas, receiving feedback, and networking. To ensure lasting impact, a comprehensive white paper summarizing key findings, discussions, and collaborative ideas generated during the workshop will be published, serving as a valuable resource to propel this exciting research direction forward. 

In particular, the objectives of this workshop are to:

• Introduce roboticists to these technologies through a combination of talks and panels  from innovators who are using performance engineering (e.g., cache-aware SIMD on the CPU), novel computing hardware (e.g., GPUs, FPGAs, custom accelerator ASICs, MCUs) and novel system designs (e.g., heterogeneous and cloud deployments) to accelerate robotics applications;

• Encourage open-ended discussion and spark new collaborations using a workshop format that includes brief small-group breakout sessions between talks and panels; and

• Collaboratively envision the future of this research direction by soliciting posters and lightning talks from students and professionals in academia and industry, as well as compiling and publishing a comprehensive white paper report.