1 paper
Haoyuan Deng, Jiebin Liu, Tengxiao Zhang +3
Pretrained robot policies provide useful action priors, but long-horizon manipulation still requires coordination between semantic reasoning and physical execution. Semantic reason…