Conditioning Matters: Training Diffusion Policies is Faster Than You Think
원문 초록
Diffusion policies have emerged as a mainstream paradigm for building vision-language-action (VLA) models. Although they demonstrate strong robot control capabilities, their training efficiency remains suboptimal. In this work, we identify a fundamental challenge in conditional diffusion policy training: when generative conditions are hard to distinguish, the training objective degenerates into modeling the marginal action distribution, a phenomenon we term loss collapse. To overcome this, we propose Cocos, a simple yet general solution that modifies the source distribution in the conditional flow matching to be condition-dependent. By anchoring the source distribution around semantics extracted from condition inputs, Cocos encourages stronger condition integration and prevents the loss collapse. We provide theoretical justification and extensive empirical results across simulation and real-world benchmarks. Our method achieves faster convergence and higher success rates than existing approaches, matching the performance of large-scale pre-trained VLAs using significantly fewer gradient steps and parameters. Cocos is lightweight, easy to implement, and compatible with diverse policy architectures, offering a general-purpose improvement to diffusion policy training.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Vision Language Action Models in Robotic Manipulation: A Systematic Review
VOTE: Vision-Language-Action Optimization with Trajectory Ensemble Voting
DreamVLA: A Vision-Language-Action Model Dreamed with Comprehensive World Knowledge
A Survey on Vision-Language-Action Models for Autonomous Driving
Parallels Between VLA Model Post-Training and Human Motor Learning: Progress, Challenges, and Trends
WorldVLA: Towards Autoregressive Action World Model
토론
이 논문과 연결된 포럼 글이 아직 없습니다.