← Leaderboard

3D Diffuser Actor

Policy
Carnegie Mellon University · Feb 16, 2024

Multi-task manipulation policy from CMU that denoises end-effector trajectories with a 3D relative-position transformer over RGB-D scene features. It is trained per benchmark (RLBench, CALVIN) rather than pretrained across embodiments. Unlike most CALVIN entries, it uses depth and camera extrinsics.

Type
Policy
Parameters
—
Backbone
–
Weights
Open
License
MIT
Commercial use
Allowed
Released
Feb 16, 2024
Robots
Franka Panda (sim), Franka
SO-101
–
LeRobot
–
Inputs
RGB · D · PC · P · L
VRAM
–

Benchmark results

1 of 13 benchmarks
CALVIN ABC→DNear saturationSimulation
#20 of 243.35 / 5, self-reported by the authors, not a leaderboard entry
Not reported on 12 benchmarks
Wrong or missing number? Report an error