Effective online adaptation of vision-language-action (VLA) models remains challenging, as sparse rewards provide weak supervision for high-dimensional autoregressive action policies. Although self-distillation can in principle provide denser training signals, we find that text-based privileged teachers conditioned on demonstrations, retrieved experiences, or high-level plans are ineffective for VLA adaptation, exposing a modality gap between symbolic guidance and low-level robot actions.
arXiv:2608. 11363v1 Announce Type: cross Abstract: A central goal in robot learning is to move beyond task-specific human data collection toward robots that improve through autonomous interaction.
By Shreyas Kowshik, Sreyas Venkataraman, Leo Wang, Niharika Pant, Max Simchowitz, Aviral Kumar
DriftOPD is a teacher‑free, rollout‑free framework that performs sequence‑level on‑policy distillation of continuous Vision‑Language‑Action (VLA) action experts. It decomposes the sequence‑level reverse‑KL divergence into a chunk‑level reverse‑KL term and a future‑potential term, optimizing them with a one‑step drifting objective and a Q‑function critic learned from offline demonstrations. Experiments on multiple VLA architectures in simulation and real‑world manipulation show that DriftOPD outperforms existing one‑step distillation baselines while matching the task success of multi‑step teacher policies.
By Youngjun Jun, Kyumin Choi, Youngmin Kim, Seonghyun Jin, Sunwoo Park, Jangho Park, Jong Chul Ye
arXiv:2603.16065v3 Announce Type: replace-cross
Abstract: Reinforcement Learning (RL) has shown strong potential for improving robotic manipulation policies, yet its practical use remains bottlenecke...
By Yanru Wu, Weiduo Yuan, Esteban Martinez Licon, Ang Qi, Vitor Guizilini, Jiageng Mao, Yue Wang
State-of-the-art vision-language-action (VLA) models such as $π_{0.5}$ exhibit strong semantic understanding, instruction following and task behavior. However, when deployed on new robots, even minor...
arXiv:2606. 31377v1 Announce Type: cross Abstract: Reinforcement learning for long-horizon robotic manipulation is often limited by sparse and delayed rewards, while manually designing dense shaping signals is costly and brittle to changes in environments and object configurations.
By Yang Yang, Bingjie Chen, Zihan Wang, Yizhe Li, Guoping Pan, Yi Cheng, Houde Liu