USC's WACL teaches robots whole-arm manipulation while the hand stays busy
ebiyik_ · x · 2026-11-05
USC researchers (Huang, Seita, Bıyık) present WACL at CoRL 2026: a hierarchical framework letting robots manipulate with their whole arm while the end-effector is occupied — holding a plate while pushing a door shut, keeping a cup upright while pressing a coffee machine. Demos are learned from bare-hand human videos with no teleop or simulation; a VLM generates safety constraints from task descriptions, no retraining needed, and the robot kinematics enter only a quasi-static surface-contact optimizer, enabling zero-shot transfer to new robots. Supported by an NVIDIA grant.
More from Embodied
- Investors back Mira Glasses: Harvard duo's camera-free AI glasses transcribe talks in 60+ languages — Scobleizer · 2026-11-05
- RoboColosseum founder: robotics lags AI because it lacks real-world benchmarks — DJiafei · 2026-11-05
- Agility Robotics goes public tomorrow via SPAC at $2.5B valuation with just $1.8M in sales — Scobleizer · 2026-11-05
- TensorWorld Launches With $105M Seed to Build Physics-Native Frontier Models — jiqizhixin · 2026-11-05
- Robotics startups will burn out before profitability — be the last mover — ___Mufasaa · 2026-11-05
- ExMR open-sources single-video full-body 3D motion capture with robot export — camenduru · 2026-11-05