TIC-VLA: A Think-in-Control Vision-Language-Action Model for Robot Navigation in Dynamic Environments

TL;DR

TIC-VLA introduces a latency-aware vision-language-action framework for robot navigation that models delayed semantic reasoning, improving real-time control in dynamic environments.

Contribution

It proposes a novel latency-aware model and training pipeline that explicitly handle reasoning delays, enhancing robot navigation performance in dynamic settings.

Findings

Outperforms prior VLA models in simulation and real-world tests.

Maintains robust real-time control despite multi-second reasoning latency.

Introduces DynaNav, a realistic simulation environment for evaluation.

Abstract

Robots in dynamic, human-centric environments must follow language instructions while maintaining real-time reactive control. Vision-language-action (VLA) models offer a promising framework, but they assume temporally aligned reasoning and control, despite semantic inference being inherently delayed relative to real-time action. We introduce Think-in-Control (TIC)-VLA, a latency-aware framework that explicitly models delayed semantic reasoning during action generation. TIC-VLA defines a delayed semantic-control interface that conditions action generation on delayed vision-language semantic states and explicit latency metadata, in addition to current observations, enabling policies to compensate for asynchronous reasoning. We further propose a latency-consistent training pipeline that injects reasoning inference delays during imitation learning and online reinforcement learning, aligning training with asynchronous deployment. To support realistic evaluation, we present DynaNav, a physics-accurate, photo-realistic simulation suite for language-guided navigation in dynamic environments. Extensive experiments in simulation and on a real robot show that TIC-VLA consistently outperforms prior VLA models while maintaining robust real-time control under multi-second reasoning latency. Project website: https://ucla-mobility.github.io/TIC-VLA/