A Neuroevolutionary Approach to Controlling Traffic Signals Based on Data from Sensor Network.

Area of Science:

• Artificial Intelligence
• Transportation Engineering
• Computer Science

Background:

• Traffic congestion is a major urban challenge.
• Current traffic signal control systems often lack real-time adaptability.
• Decentralized control offers potential for improved traffic management.

Purpose of the Study:

• To develop an artificial neural network ensemble for decentralized traffic signal control.
• To enhance traffic signal efficiency using real-time sensor data.
• To optimize traffic flow by reducing vehicle delays and travel times.

Main Methods:

• Implemented a decentralized approach with independent intersection control.
• Utilized an artificial neural network ensemble combining intersection-specific and fully connected networks.
• Employed a neuroevolution strategy for optimizing the neural network ensemble configuration.
• Conducted extensive simulation experiments comparing the proposed solution with state-of-the-art algorithms.

Main Results:

• The proposed artificial neural network ensemble demonstrated superior performance.
• Significant reductions in vehicle delay were observed.
• Substantial decreases in overall travel time were achieved.
• Notable increases in average vehicle velocity were recorded.

Conclusions:

• The developed artificial neural network ensemble is an effective solution for decentralized traffic signal control.
• The system offers a promising approach to improving urban traffic efficiency.
• This method outperforms existing decentralized traffic control algorithms.