RPL humanoid perceptive locomotion code released, accepted to CoRL 2026
GuanyaShi · x · 2026-10-21
The RPL team (authors include Pieter Abbeel, Karen Liu, Rocky Duan, Guanya Shi) released full code for RPL: Learning Robust Humanoid Perceptive Locomotion on Challenging Terrains, accepted to CoRL 2026.
- A unified policy handles slopes, stairs, stepping stones, multiple walking directions, and payload disturbances — even when different cameras see different terrains
- Trained in simulation, validated with long-horizon real-world runs; its perceptive features already adopted by parallel works like LadderMan and PHP
- Release covers expert training, multi-depth distillation, and sim2sim; you can drive it in Mujoco with an Xbox controller
More from Embodied
- Basalt monocular-inertial VIO ported to WASM: SLAM now runs live in mobile browsers — rsasaki0109 · 2026-10-21
- DIY voice-enabled AI companion device with 3D-printed shell and pet avatar — shekitup · 2026-10-21
- RoboARC shows robots get no free retries—unlike coding agents that can just rerun tests — ManlingLi_ · 2026-10-21
- All 20 robot arms failed in overcast weather at demo day, sparking low-level control debate — chris_j_paxton · 2026-10-21
- 'They still fail the burrito test': the case for putting AI models in bodies — generativist · 2026-10-21
- Three JEPA variants, one question: how well must a machine predict to plan? — TheTuringPost · 2026-10-21