SafeLight: A Reinforcement Learning Method toward Collision-Free Traffic Signal Control

Abstract

Traffic signal control is safety-critical for our daily life. Roughly one-quarter of road accidents in the U.S. happen at intersections due to problematic signal timing, urging the development of safety-oriented intersection control. However, existing studies on adaptive traffic signal control using reinforcement learning technologies have focused mainly on minimizing traffic delay but neglecting the potential exposure to unsafe conditions. We, for the first time, incorporate road safety standards as enforcement to ensure the safety of existing reinforcement learning methods, aiming toward operating intersections with zero collisions. We have proposed a safety-enhanced residual reinforcement learning method (SafeLight) and employed multiple optimization techniques, such as multi-objective loss function and reward shaping for better knowledge integration. Extensive experiments are conducted using both synthetic and real-world benchmark datasets. Results show that our method can significantly reduce collisions while increasing traffic mobility.

Prompt to Transfer: Sim-to-Real Transfer for Traffic Signal Control with Prompt Learning

“…Traffic Signal Control (TSC) is a critical task aimed at improving transportation efficiency and alleviating congestion in urban areas (Wei et al 2021). Reinforcement Learning (RL) methods have shown promising results in tackling TSC challenges through trial and error in simulators (Ghanadbashi and Golpayegani 2022;Mei et al 2023;Noaeen et al 2022;Ducrocq and Farhi 2023;Zang et al 2020;Wu et al 2020;Haydari and Yılmaz 2020;Du et al 2023;Vlachogiannis et al 2023), bringing hope for solving cities' traffic congestion issues. While simulation is a valuable tool for control tasks in the real world with low cost, notable performance gaps arise when deploying simulator-trained policies to real-world environments (Da et al 2023b,c), mainly due to differences in system dynamics between training simulators and the actual road conditions.…”