Vision-Language-Action models have shown great promise for autonomous driving, yet they suffer from degraded perception after unfreezing the visual encoder and struggle with accumulated instability in long-term planning. To address these challenges, we propose EvoDriveVLA-a novel collaborative perception-planning distillation framework that integrates self-anchored perceptual constraints and future-informed trajectory optimization. Specifically, self-anchored visual distillation leverages self-anchor teacher to deliver visual anchoring constraints, regularizing student representations via trajectory-guided key-region awareness. In parallel, future-informed trajectory distillation employs a future-aware oracle teacher with coarse-to-fine trajectory refinement and Monte Carlo dropout sampling to synthesize reasoning trajectories that model future evolutions, enabling the student model to internalize the future-aware insights of the teacher. EvoDriveVLA achieves SOTA performance in nuScenes open-loop evaluation and significantly enhances performance in NAVSIM closed-loop evaluation. Our code is available at: https://github.com/hey-cjj/EvoDriveVLA.
@article{arxiv.2603.09465,
title = {EvoDriveVLA: Evolving Driving VLA Models via Collaborative Perception-Planning Distillation},
author = {Jiajun Cao and Xiaoan Zhang and Xiaobao Wei and Liyuqiu Huang and Zijian Wang and Hanzhen Zhang and Zhengyu Jia and Wei Mao and Hao Wang and Xianming Liu and Shuchang Zhou and Yang Wang and Shanghang Zhang},
journal= {arXiv preprint arXiv:2603.09465},
year = {2026}
}