When you are late for work or just trying to meet up for that appointment, that extra drive time you spend in traffic can be more than a little frustrating. This frustration gets worse as time goes by. The blaring horns, the screaming, and the cursing can be a little too much to handle. Statistics show that in 2019, the average UK resident wasted around 115 hours and £894 ($1,115) in fuel waste and lost income yearly due to traffic jams across the UK’s urban areas. Ironically, we have traffic lights and road regulations aimed at reducing traffic jams, yet a major cause of traffic jams comes from the existing traffic lights. Studies show that ineffective traffic signal timing accounts for the main cause of traffic jams experienced daily.
Read Also: Air Pollution in Cities Responsible for Nearly Two Million Excess Deaths Worldwide Annually
Traffic Jam
However, researchers at Aston University have created a new artificial intelligence AI system that uses the ‘principle of learned behaviors’ to ensure a continuous traffic flow. It is hoped that using this system, long queues at traffic lights could become a thing of the past.
How does the AI traffic light system work?
Imagine a traffic light system capable of adapting rapidly to the situation in real-time and receiving feedback about the traffic situation. Using live camera footage, the AI traffic light reads the situation and adapts the lights to compensate, keeping the traffic flowing and reducing congestion. The AI system also receives feedback on the effectiveness of its work, and using deep reinforcement learning, it learns to make the necessary adjustments.
This system was first tested virtually using a state-of-the-art photo-realistic traffic stimulator known as Traffic 3D. The AI program was exposed to different traffic and weather scenarios, using the stimulator, thereby teaching it to handle these situations. This simulation worked on a reward system where the system got a positive reward each time it got a car through the traffic, but got a negative reward each time a car had to wait in line. Interestingly, the system does not require any additional input to run, as the only thing the researchers did was control the reward system of the simulator. The researchers also disclosed that the reward system could be altered to allow the system to prioritize certain emergency vehicles, such as ambulances.
Amazingly, when the system was later tested on a physical junction, it subsequently adapted to the real traffic intersections despite being trained on a different simulation, therefore showing the ability of the system to adapt to real-life situations.
Importance of the AI system: a world with no traffic jams
Traffic congestion is a worldwide problem, and this research by scientists at Aston University hopes to solve this problem. The AI system would ensure reduced drive time, increased ease to navigate the roads, and above all, reduce the frustration associated with waiting in traffic jams.
Read Also: Cameras and Sensors to Monitor Passenger’s Health in Self-Driving Cars
Conclusion
The AI traffic system is a remarkable invention that would surely reduce traffic congestion considerably worldwide. Researchers hope to begin testing their systems on real roads this year.
References
Fully-Autonomous, Vision-based Traffic Signal Control: from Simulation to Reality
