3 papers indexed
Jinbang Huang, Yuanzhao Hu, Zhiyuan Li, Ran Qi, Yixin Xiao, Zhanguang Zhang, et al.
Long-horizon robotic tasks require diverse capabilities that no single policy can reliably provide. Heterogeneous policies offer complementary strengths, but orchestrating them requires reasoning over uncertain capability boundaries and cross-policy distribution mismatch, which a…
Lingfeng Zhang, Zhanguang Zhang, Liheng Ma, Tongtong Cao, Yingxue Zhang
End-to-end vision-language navigation (VLN) with causal vision-language models maps instructions and egocentric observations directly to actions, but standard behavior cloning supervises only the next action and does not explicitly encourage the policy state to be predictive of f…
Xinyu Shao, Keru Zhou, Guowei Huang, Yajun Gao, Tongtong Cao, Xiu Li
Learning manipulation from few demonstrations requires visual priors that capture not only where to interact, but also how the interaction should begin; static priors such as segmentation masks encode only the former. We present KAM-WM, a framework that extracts a coarse directio…