TY - RPRT TI - APS-VLA: Action-Guided and Future-Aware Token Pruning for Efficient Vision-Language-Action Models AU - Xiaoyang Xing AU - Chun-Mei Feng AU - Anthony Cohn AU - Xiaoliang Gong AU - Asoke Nandi PY - 2026 UR - https://searcharxiv.com/abs/2611.13030 ID - 2611.13030 N1 - Source identifier: 2611.13030 ER -