Q-Net: Queue Length Estimation via Kalman-based Neural Networks
Researchers have developed Q-Net, a novel framework for estimating traffic queue lengths at signalized intersections. This AI-augmented Kalman filter integrates data from loop detectors and floating car data, addressing challenges like differing data resolutions and traffic conservation violations. Evaluations in Rotterdam demonstrated Q-Net's superior performance compared to baseline methods, offering accurate tracking of queue dynamics without expensive sensing infrastructure. AI
IMPACT Introduces a novel AI-driven method for traffic management, potentially reducing the need for costly sensor infrastructure.