Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

RISE: 3D Perception Makes Real-World Robot Imitation Simple and Effective

About

Precise robot manipulations require rich spatial information in imitation learning. Image-based policies model object positions from fixed cameras, which are sensitive to camera view changes. Policies utilizing 3D point clouds usually predict keyframes rather than continuous actions, posing difficulty in dynamic and contact-rich scenarios. To utilize 3D perception efficiently, we present RISE, an end-to-end baseline for real-world imitation learning, which predicts continuous actions directly from single-view point clouds. It compresses the point cloud to tokens with a sparse 3D encoder. After adding sparse positional encoding, the tokens are featurized using a transformer. Finally, the features are decoded into robot actions by a diffusion head. Trained with 50 demonstrations for each real-world task, RISE surpasses currently representative 2D and 3D policies by a large margin, showcasing significant advantages in both accuracy and efficiency. Experiments also demonstrate that RISE is more general and robust to environmental change compared with previous baselines. Project website: rise-policy.github.io.

Chenxi Wang, Hongjie Fang, Hao-Shu Fang, Cewu Lu• 2024

Related benchmarks

TaskDatasetResultRank
One-Move manipulationOne-Move
Success Rate65
12
Three-Scoop manipulationThree-Scoop
Success Rate (SR)15
12
Swap-Easy manipulationSwap-Easy
SR15
12
Add-Salt manipulationAdd-Salt
SR45
12
Swap-Hard manipulationSwap-Hard
SR10
9
Robotic InsertionCobot Mobile ALOHA In-distribution (train)
Task 1 Success Rate20
5
Open OvenReal-world dexterous manipulation
Hook Success Rate100
4
Open jarReal-world dexterous manipulation
Hook Success Rate80
4
Pull TissueReal-world dexterous manipulation
Grasp Success Rate75
4
Toast BreadReal-world dexterous manipulation
Grasp Success Rate80
4
Showing 10 of 19 rows

Other info

Follow for update