Back on the road with Deep Reinforcement Learning at the Traffic lights


Back on the road with Deep Reinforcement Learning at the Traffic lights

When you’re next waiting at the traffic lights, you may be wishing you were in China, thanks to the research being done by Xiaoyuan Liang, Xusheng Du, Guiling Wang, and Zhu Han. They have been working on deep reinforcement learning for traffic.

Decreasing the pollution created by and energy wasted by vehicles waiting at traffic lights could be a real game changer. Even the smallest gain of 1 or 2 percent could see serious changes to the environment.

While there have been previous studies into using Deep Reinforcement Learning to control traffic signals, Xiaoyuan Liang and teams research takes it further with a Convolutional Neural network effectively on top, creating a proposed whole network called Double Dueling Deep Q Network (3DQN)

And the results were good. They calculated the average cumulative waiting time as the measurable result. And in both rush hour and normal traffic rate situations the system reduced the waiting time by over 20%.

Next time you are at a council meeting, maybe you could raise the new options available to the county planners, a 20% cut in pollution is surely worth investment.


Make Contact

Click here to open up the message window or call us on 01344 567990

Want to know more ?

Please enter your name and email address and/or contact number

We use cookies to provide you with the best possible experience in your interactions on our website

You agree to our use of cookies on your device by continuing to use our website

I understand