Enabling Embodied AI with Open-Source Solutions from AMD

Abstract:

AI-enabled robots are rapidly becoming a reality, propelled by advancements in Embodied AI, software, and semiconductor technologies.
 
The talk will explore how evolving Embodied AI models are improving robots’ ability to perceive, learn, and interact in human environments. It will also highlight AMD Research's role in supporting open-source projects with complete solutions, focusing on 2 examples: the LeRobot ecosystem and Unitree robots. For attendees interested in hardware, these solutions will highlight the combination of AI PCs with traditional FPGA SoCs to realize embodied AI requirements.  Finally, a preview of our ROSCon’25 workshop agenda, featuring cloud-hosted AI PCs, will be shown.
 
The primary aim of this talk is to broadly present AMD Research's robotics work to encourage collaboration with ETH Zürich.

Bio:

Ken O'Brien is a senior member of technical staff within the AMD Research and Advanced Development Group (RAD), in Dublin, Ireland. He has worked in the areas of reduced precision machine learning, bioinformatics, network acceleration for AI/HPC workloads and systems software for FPGA/GPU disaggregated computing. He is currently researching AI driven robotics.  He holds a PhD in computer science from UCD, Dublin.

Graham Schelle is a fellow at AMD within the Research and Advanced Development Group (RAD) based in Colorado, USA.  He leads a small team within AMD Research looking at productivity abstractions across complex software and hardware solutions.  Additionally, he is the lead maintainer of both the PYNQ and the RFSoC-PYNQ open-source projects.  Graham is also a member of AMD’s Office of Open Source.