Controlling traffic lights to optimize vehicle flow
One of the key challenges for cities is to optimize traffic flows on road networks. There are numerous benefits in reducing traffic congestions, including but not limited to:
- Reducing the time and energy wasted in traffic
- Saving on gas and resulting exhaust emissions
- Increasing vehicle and road lifetime
- Decreasing number of accidents
There has been already a lot of research going in this area; but recently, RL has emerged as a competitive alternative to traditional control approaches. So, in this section, we optimize the traffic flow at a road network by controlling the traffic light behavior using multi-agent reinforcement learning. To this end, we use the Flow framework, which is an open-source library for RL and control experiments on realistic traffic microsimulations.
Introducing Flow
Transportation research significantly relies on simulation software, such as SUMO and Aimsun, for topics such...
