Tsinghua Releases TACO Dataset for Bimanual Robot Tool Manipulation
YuXiang_IRVL · x · 2026-08-16
Tsinghua University and collaborators released TACO, a large-scale dataset for bimanual hand-object manipulation.
- Content: Covers extensive tool-action-object combinations in real-world scenarios, featuring 2.5K motion sequences with both third-person and egocentric views.
- Specs: Provides precise hand-object 3D mesh annotations and action labels, supporting generalization to unseen object geometries.
- Use Cases: Benchmarks generalizable research topics such as action recognition, motion forecasting, and cooperative grasp synthesis.
- Availability: Dataset and code are open source.
Related event: Tsinghua Releases TACO Dataset for Dual-Arm Robots(2 posts)→
More from Embodied
- Tesla says new Roadster's aero package is driven by Optimus actuators, nearing F1 downforce — yunta_tsai · 2026-10-16
- Tesla Roadster event end sizzle reel surfaces online — banteg · 2026-10-16
- Up-close photos reveal the Roadster's cold-gas thruster tank and nozzles — PTrubey · 2026-10-16
- Real-time collaborative mapping across phones and robots on the Auki network — broodsugar · 2026-10-16
- Tesla Roadster does 0-60 mph in 0.81 seconds — not a typo — Polymarket · 2026-10-16
- Tesla Roadster stuns live demo: car appears to fly, then drives upside down — XFreeze · 2026-10-16