Understanding Multimodal Failure in Action-Chunking Behavioral Cloning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910246135398400 |
|---|---|
| author | Mazza, Lorenzo Datres, Massimiliano Rodriguez, Ariel Bodenstedt, Sebastian Kutyniok, Gitta Speidel, Stefanie |
| author_facet | Mazza, Lorenzo Datres, Massimiliano Rodriguez, Ariel Bodenstedt, Sebastian Kutyniok, Gitta Speidel, Stefanie |
| contents | Behavioral cloning becomes difficult when the same observation admits several valid actions. We study this problem for action-chunking policies and show that different multimodal parameterizations fail in different ways. For latent-variable policies, posterior-prior regularization makes deployment-time sampling more reliable, but excessive regularization removes the action-conditioned information needed to distinguish demonstrated modes. Reducing this regularization can preserve mode information, but then success depends on whether the prior covers the relevant latent regions. For action-space generative policies, multimodality is constrained by the smoothness of the base-to-action transport: a map with small Lipschitz constant cannot assign substantial probability to many well-separated modes. Covering many modes therefore requires either sharp transitions in base space or off-support bridge regions in action space. Experiments on synthetic multimodal tasks and robotic simulation benchmarks support these mechanisms. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_22493 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Understanding Multimodal Failure in Action-Chunking Behavioral Cloning Mazza, Lorenzo Datres, Massimiliano Rodriguez, Ariel Bodenstedt, Sebastian Kutyniok, Gitta Speidel, Stefanie Machine Learning Artificial Intelligence Robotics Behavioral cloning becomes difficult when the same observation admits several valid actions. We study this problem for action-chunking policies and show that different multimodal parameterizations fail in different ways. For latent-variable policies, posterior-prior regularization makes deployment-time sampling more reliable, but excessive regularization removes the action-conditioned information needed to distinguish demonstrated modes. Reducing this regularization can preserve mode information, but then success depends on whether the prior covers the relevant latent regions. For action-space generative policies, multimodality is constrained by the smoothness of the base-to-action transport: a map with small Lipschitz constant cannot assign substantial probability to many well-separated modes. Covering many modes therefore requires either sharp transitions in base space or off-support bridge regions in action space. Experiments on synthetic multimodal tasks and robotic simulation benchmarks support these mechanisms. |
| title | Understanding Multimodal Failure in Action-Chunking Behavioral Cloning |
| topic | Machine Learning Artificial Intelligence Robotics |
| url | https://arxiv.org/abs/2605.22493 |