We address the traffic light control problem for a single intersection byviewing it as a stochastic hybrid system and developing a Stochastic Flow Model(SFM) for it. We adopt a quasi-dynamic control policy based on partial stateinformation defined by detecting whether vehicle backlog is above or below acertain threshold, without the need to observe an exact vehicle count. Thepolicy is parameterized by green and red cycle lengths which depend on thispartial state information. Using Infinitesimal Perturbation Analysis (IPA), wederive online gradient estimators of an average traffic congestion metric withrespect to these controllable green and red cycle lengths when the vehiclebacklog is above or below the threshold. The estimators are used to iterativelyadjust light cycle lengths so as to improve performance and, in conjunctionwith a standard gradient-based algorithm, to seek optimal values which adapt tochanging traffic conditions. Simulation results are included to illustrate theapproach and quantify the benefits of quasidynamic traffic light control overearlier static approaches.
展开▼