Fudan and Fysics AI unveil UniWAM, hitting 80% success on real-robot mobile manipulation
jiqizhixin · x · 2026-11-17
Fysics AI and Fudan University introduced UniWAM, a unified world-action model for navigation and manipulation, detailed in an arXiv report.
- Reaches 80% success across four real-robot mobile-manipulation tasks, 17.5 points above the π₀.₅ baseline
- A Manipulation Anchor Pose (MAP) specifies where the robot should stop and face before acting, with generated supervision teaching approach trajectories
- Navigation and manipulation use separate action interfaces but share video and action Transformer backbones; over 1.5M MAP-labeled episodes scale navigation supervision
- On MAP-Bench: 9.85 cm average position error and 2.30° heading error
More from Embodied
- XREAL Aura review: the spatial computer Apple should have made — Scobleizer · 2026-11-17
- ScaleVLA scales VLA models to 552B parameters: 10 demos beat 2B backbone trained on 50 — GalaxyGeneralRobotics · 2026-11-17
- Amazon's next delivery runner has four legs: RIVR robot joins the last-mile fleet — TinfoilTricorn · 2026-11-17
- ChipAgents CEO joins NYSE interview on chip AI expansion and chip superintelligence vision — WilliamWangNLP · 2026-11-17
- Nvidia Robotics Chief Says OpenAI's Astra Makes Robotic Training Data 100x Faster — pstAsiatech · 2026-11-17
- Ant Group uses action-free egocentric human videos to boost robot policy learning — antgroup · 2026-11-17