English

DeCoR: Design and Control Co-Optimization for Urban Streets Using Reinforcement Learning

Machine Learning 2026-05-21 v1 Artificial Intelligence

Abstract

Modern vision systems can detect, track, and forecast urban actors at scale, yet translating perception outputs to urban design remains limited. We introduce DeCoR, a two-stage reinforcement learning framework that leverages flow observations to co-optimize crosswalk layout and network-level signal control. The design stage encodes the pedestrian network as a graph and learns a generative policy that parameterizes a Gaussian mixture model over crosswalk location and width, from which new crosswalks are sampled. For each layout, a shared control policy learns adaptive signal timings to minimize joint pedestrian and vehicle delay. On a 750 m real-world urban corridor with demand sensed from video and Wi-Fi logs, DeCoR learns a layout that reduces pedestrian arrival time to their nearest crosswalk by 23% while using fewer crosswalks than existing configurations. On the control side, DeCoR reduces pedestrian and vehicle wait time by 79% and 65%, respectively, relative to fixed-time signalization. Further, the control policy generalizes to demands outside of training and is robust to layout changes without retraining.

Keywords

Cite

@article{arxiv.2605.21311,
  title  = {DeCoR: Design and Control Co-Optimization for Urban Streets Using Reinforcement Learning},
  author = {Bibek Poudel and Lei Zhu and Kevin Heaslip and Sai Swaminathan and Weizi Li},
  journal= {arXiv preprint arXiv:2605.21311},
  year   = {2026}
}

Comments

22 pages, 8 figures