MIT · robot AI model
T-WLA
United States · 2026
Runs on Linker Hand L6 by LinkerbotSource and note: Linker Hand L6D arxiv.org/html/2610.28814v2 · Retrieved 19 November 2026
Sources retrieved: 19 November 2026 (A, B, C, D)
- Model type
- World action modelSource and note: Model typeB arxiv.org/abs/2610.28814 · Retrieved 19 November 2026Note. “We introduce T-WLA, a world-language-action architecture that uses shared visual memory to guide both action generation and tactile correction.”
- What it controls
- Arms and handsSource and note: What it controlsB arxiv.org/abs/2610.28814 · Retrieved 19 November 2026Note. “We evaluate T-WLA on tactile-conditioned hand-motion prediction using EgoTouch and Tachin, and on four real-world dexterous manipulation tasks.”
- Inputs
- Camera Language Robot state TouchSource and note: InputsA github.com/kaichen-z/TWLA · Retrieved 19 November 2026Note. “At each RGB update, retrieve past visual information and combine it with the current observation, instruction, state, and field of view to produce visual memory and a base action chunk.” “The same memory is cached for a lightweight tactile editor, which combines it with the active action chunk and incoming touch to predict residual corrections to unexecuted actions without rerunning the visual branch.”
- Access
- Open weights (download)Source and note: AccessA github.com/kaichen-z/TWLA · Retrieved 19 November 2026Note. “Visual-policy, RGB-world-head, and tactile-encoder weights are linked above.”

Specifications
6 of 14 stated 8 not stated
5 fields hidden 3 not-stated fields with a note stay ·
How to read the figuresShow guideClose
- not statedNo source
- 0Reported as zero
- ASource
Select a source marker for source details and notes.
Release1/1 disclosed
Input and output3/4 disclosed
Size and speed1/2 disclosed
Where it runs and what it is built on0/2 disclosed
Openness1/3 disclosed
The developer’s own claims0/2 disclosed
Notes
Developed by researchers from MIT, Harvard, UC Berkeley, and UC Davis.
Described by the authors as “We introduce T-WLA, a world-language-action architecture that uses shared visual memory to guide both action generation and tactile correction.”
Sources
One line per source, not per figure: a source has one retrieval date.
- Agithub.com/kaichen-z/TWLARetrieved 19 November 2026
- Barxiv.org/abs/2610.28814Retrieved 19 November 2026
- Ct-wla.github.ioMain source. Retrieved 19 November 2026
- Darxiv.org/html/2610.28814v2Retrieved 19 November 2026



