Smart Signal – Adaptive Traffic Signal Control using Reinforcement Learning and Object Detection

Abstract

In order to solve the problems that reduce traffic efficiency in the road network, which is caused by overlarge traffic demand of urban regions at peak time, where traditional traffic light systems are unable to handle, we are proposing a working model of smart adaptive traffic signal control system based on single-agent reinforcement learning algorithm. Our system uses object detection algorithm to sense real-time traffic scenario, which is used by the reinforcement learning algorithm to compute optimal signal timing for the given scenario. Simulation results show that our model has good practicability and is substantially cheaper than the systems currently in use.