Modeling What Changes: Sparse, Residual World Models for Object-Centric Manipulation
技术动态AI 88
来源：arXiv cs.AI发布时间待核实
arXiv:2609.02046v1 Announce Type: cross Abstract: Monolithic world models predict the entire next state at every step, spending capacity re-predicting the static majority of a scene and injecting error into it. We ask whether explicitly modeling change (a per-object change gate plus a residual delta head that perturbs only the objects the gate flags) is a more effective and interpretable bias for physical prediction and control.