The paper “TIC-VLA: A Think-in-Control Vision-Language-Action Model for Robot Navigation in Dynamic Environments” is accepted for ICML 2026! Checkout the paper, code, and website.