Microsoft Research challenges the assumption that physical AI inference must run exclusively on onboard robot GPUs, demonstrating that offloading to edge or cloud infrastructure significantly enhances mobile manipulation workloads. Their systematic study of robotics workloads reveals that relying on onboard compute limits performance, battery life, and scalability.
MuJoCo Warp (MJWarp), built on NVIDIA Warp, allows compatible MuJoCo models to run at GPU scale, enabling the advancement of hundreds or thousands of independent simulation worlds in a single call. This architecture shifts the focus from minimizing latency for one environment to improving aggregate throughput, which favors reinforcement learning and large-scale sampling.
The article introduces Human2Any, a method for human-to-robot transfer that utilizes constraint-aware compositional planning. No further details or body text are provided in the source.
The authors present Show-Harness, an Embodied Harness that allows vision-language models (VLMs) to control robots through a compact semantic interface linking intent to action. This system exposes discrete semantic action units for VLM reasoning and uses embodiment-specific interpreters to ground them into local robot actions.
The author has released Calibra, a tool designed to identify poor demonstrations in robotics imitation learning datasets before they consume GPU resources.