DP3 (3D Diffusion Policy)

Visuomotor diffusion policy conditioned on a compact 3D point-cloud representation (RSS 2024), from Shanghai Qi Zhi Institute, SJTU, Tsinghua IIIS and Shanghai AI Lab. It is a single-task baseline on the RoboTwin 2.0 leaderboard, where the board marks it as needing depth.

Shanghai Qi Zhi Institute et al. · Mar 6, 2024 · PolicyReport an error
30.1%RoboTwin 2 · #29 of 33
1Benchmarks reported
Mar 6, 2024Released
OpenWeights

Benchmark results1

RoboTwin 2.0Near saturationSimulation
#29 of 3330.1%
Not reported on 14 benchmarks
Wrong or missing number? Report an error