GraSP-VLA: Graph-based Symbolic Action Representation for Long-Horizon Planning with VLA Policies

arXiv · 2025-11-06 · Low-Level Learning-Based Action Modelling / Input Modelling / 2D Structured Vision Language Action Models