Vision language action model improves traffic signal control decisions
VLALight: A Vision-Language-Action Model for Traffic Signal Control
Artificial Intelligence
Summary
Traffic lights help manage city traffic but deciding when to change them can be tricky. The authors designed a computer model called VLALight that watches videos from cameras at intersections and decides the best traffic light changes directly from these images. Their system learns to think carefully when needed and works together across intersections to make better overall traffic flow. Tests using real city traffic data show VLALight makes smarter traffic signals than previous methods.
What this means in practice
- •For urban traffic engineers: Deploy coordinated traffic signal systems that use real-time camera feeds for improved congestion management across multiple intersections.
- •For smart city platform developers: Integrate multi-view visual traffic analysis with adaptive signal control to enhance city-wide traffic efficiency with learned reasoning modes.
Authors
Pan Zhang, Siqi Lai, Kemu Dong, Hao Liu
Abstract
Traffic signal control (TSC) is essential for improving urban mobility and reducing congestion. Although roadside cameras are widely deployed at signalized intersections and provide rich visual observations of evolving traffic, existing TSC methods typically rely on manually engineered traffic states or separate perception modules, creating a gap between physical observations and control decisions. We present VLALight, the first vision-language-action (VLA) model for end-to-end traffic signal control from multi-view roadside videos. VLALight directly maps visual observations to coordinated signal actions through multi-target spatiotemporal traffic reasoning and topology-aware cooperative perception across intersections. To establish this capability, we develop a two-stage supervised cold-start training strategy for visual traffic understanding and signal decision-making, followed by cooperative agentic reinforcement learning that jointly optimizes local control and network-wide traffic efficiency. Furthermore, VLALight introduces adaptive fast and slow reasoning modes, enabling the policy to allocate deeper reasoning only when additional deliberation provides sufficient control benefits. Through balanced mode-aware rollouts and relative advantage optimization, VLALight learns to trade off decision quality and inference cost. Extensive experiments on seven real-world traffic-flow datasets across three urban networks demonstrate that VLALight consistently outperforms transportation-based, RL-based, and LLM/VLM-based baselines. Ablation studies validate the effectiveness of cooperative perception, network-level optimization, and adaptive reasoning. These results demonstrate the potential of VLA models for real-world physical traffic control. Our project is available at https://github.com/usail-hkust/VLALight.git.