ETH Zurich teaches a robotic hand to walk on its fingers with reinforcement learning
MIT_CSAIL · x · 2026-10-20
MIT CSAIL highlighted an ETH Zurich paper, Fingers as Legs, showing an anthropomorphic robotic hand that uses the same fingers for locomotion, self-support, and manipulation—no separate walking mechanism needed.
- Reinforcement learning keeps the original finger design and position controllers, with reward shaping adapted to unequal fingers; the simulator is calibrated from hardware measurements
- In simulation, the hand moves faster than with tuned quadruped rewards
- On hardware, task-specific policies enable untethered crawling, steering, and fall recovery, plus keyboard typing without vision and object pushing via overhead visual feedback
- Onboard power and compute make it a compact, self-contained mobile manipulator
More from Embodied
- Tesla shows Optimus V3 render in investor letter, Musk touts an 'Amazing Abundance' era — JOBhakdi · 2026-10-20
- UCSD PhD Nicklas Hansen joins NYU as postdoc with LeCun and Pinto on world models + robots — LerrelPinto · 2026-10-20
- AI2's MolmoB0T: Zero-Shot Real-World Manipulation from 1.7M Simulated Trajectories — chris_j_paxton · 2026-10-20
- Chinese humanoid robot shows off pole dancing skills in viral demo — LexiLove · 2026-10-20
- F-22 controls a CCA drone for the first time, Anduril and USAF hit milestone — BrettKrieger12 · 2026-10-20
- Ropedia W0 shows 10k hours of human video yields a robot learning scaling law — liuziwei7 · 2026-10-20