Malik's three levels of manipulation: LLMs live on top, robots stall on forces and torques
le_james94 · x · 2026-10-13
First post of lejames94's thread: Jitendra Malik's ETH lecture splits robot manipulation into three levels, each named for a thinker — goals and plans (Aristotle: "I want to make an omelet"; language models live here), trajectories (Euclid: the same pickup verb looks nothing alike on a sugar cube vs a heavy tray), and forces and torques (Newton: grip force, friction, gravity, contact — the hard part).
A language model can write the omelet recipe; robots stall on the bottom two levels, needing to crack the egg 99.99% of the time across locations, objects, and clutter. His slide's status board: navigation close to solved (GOAT: 83% across 9 unseen homes), locomotion moving fast (VideoMimic: human video → sim → real), manipulation wide open (needs multi-fingered hands and touch).
More from Embodied
- RT-SAFE: Agents Hit 94.1% Task Success but Only 0.7% Safe Success — Lianhuiq · 2026-10-13
- Tesla: The Only Robot Cars You Can Actually Buy — yunta_tsai · 2026-10-13
- NVIDIA takes RTX Spark laptop launch events global, with a Jensen-signed ASUS ProArt P14 up for grabs — NVIDIAAI · 2026-10-13
- Actuator speed, not just AI, is what's holding robots back from snappy actions — ayushthakur0 · 2026-10-13
- Stellantis demos Fiat 500e with Wayve's self-driving AI on Turin streets — alexgkendall · 2026-10-13
- Wayve CEO Stages at Wave to Push Autonomy into Fiat and Maserati — alexgkendall · 2026-10-13