Prox Industries Advises Mitsubishi Electric on Multi-Robot Simulation
Mitsubishi Electric is building the simulation and coordination layer for Physical AI, and it brought in an outside specialist to advise on multi-robot verification rather than handling that piece internally.
Reporting from 1 source: ASCII.jp.
Prox Industries provided technical advisory support to Mitsubishi Electric for simulation-based verification of coordinated operation among multiple robots, part of Mitsubishi Electric's Physical AI research. Prox advised on building simulation environments with NVIDIA Omniverse and Isaac Sim, configuring reinforcement learning via Isaac Lab, linking vision and language models to robot actions, and verifying information sharing between robots. Subjects included quadruped walking, object transport, and robotic arm grasping.
The work covered three areas. Prox Industries advised on constructing robot models and verification environments and on reinforcement learning configurations in Isaac Lab, using quadruped walking and object transport, plus robot-arm grasping and moving, as test subjects. It also advised on linking camera-based sensing and vision-language models to robot operation, and on interpreting language instructions to select a transport destination. A third track covered mechanisms for robots to communicate, share information, and link their operations, with Prox organizing technical issues that surfaced during verification and advising on improvement policies.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.