Abstract: Vision-based imitation learning has shown promising capabilities of endowing robots with various motion skills given visual observation. However, current visuomotor policies fail to adapt to drastic changes in their visual observations. We present Perception Stitching that enables strong zero-shot adaptation to large visual changes by directly stitching novel combinations of visual encoders. Our key idea is to enforce modularity of visual encoders by aligning the latent visual features among different visuomotor policies. Our method disentangles the perceptual knowledge with the downstream motion skills and allows the reuse of the visual encoders by directly stitching them to a policy network trained with partially different visual conditions. We evaluate our method in various simulated and real-world manipulation tasks. While baseline methods failed at all attempts, our method could achieve zero-shot success in real-world visuomotor tasks. Our quantitative and qualitative analysis of the learned features of the policy network provides more insights into the high performance of our proposed method.
Submission Length: Regular submission (no more than 12 pages of main content)
Changes Since Last Submission: Make the file to have the camera-ready format. Put names on the previously anonymous video. Release the code on github.
Video: https://www.youtube.com/watch?v=H6SD9Tcvhrg
Code: https://github.com/generalroboticslab/PerceptionStitching
Supplementary Material: zip
Assigned Action Editor: ~Emmanuel_Bengio1
Submission Number: 3138
Loading