Accepted: Active Learning-Enhanced Traffic Signal Control in TRC

I am pleased to announce that our paper “Active learning-enhanced constrained Bayesian optimization for multi-modal and multi-objective traffic signal control” has been accepted for publication in Transportation Research Part C: Emerging Technologies.

This collaborative research with Dr Yunhai Gong, Christoffer Riis, Dr Shaopeng Zhong, Tao Wang, Filipe Rodrigues, and Carlos Lima Azevedo develops four constrained multi-objective Bayesian optimisation methods for traffic signal control.

The framework considers private cars, buses, and bicycles alongside three objectives: traffic efficiency, safety, and equity. Two of the proposed methods combine fully Bayesian Gaussian processes with active learning. Tests at a Copenhagen intersection showed improvements over NSGA-III and COMBOO, while application to a real-world 3 × 3 urban network demonstrated scalability to a larger multi-intersection setting.

Read the paper on ScienceDirect · View the DOI record · Read the accessible research summary