DI-smartcross Documentation

DI-smartcross

Decision Intelligence Platform for Traffic Crossing Signal Control.

Last updated on 2022.08.09


DI-smartcross is an open-source application platform under OpenDILab. DI-smartcross uses Reinforcement Learning in precise control of traffic crossing signals in order to optimize transportation time cost by coordinating vehicles’ movements at crosses. DI-smartcross applies training & evaluation for various RL policies using DI-engine in provided road nets. DI-smartcross supports SUMO and CityFlow simulators to enable traffic flow simulation with different granularity.

Main Features

  • Design easy-to-use crossing signal control environments, with various State, Action, and Reward options.

  • Build a variety of road networks of different scales, ideal or from the real world.

  • Adapting several Reinforcement Learning strategies using DI-engine, including discrete or continuous space, multi-agent etc.

Content

Installation

Quick Start

RL Environments

FAQ