Lateral and Longitudinal Trajectory Optimisation Method for a Mixed Fleet of Connected and Automated Vehicles Passing Through Multiple Continuous Intersections

The efficiency of signalised intersections affects the overall performance of urban transportation systems. Despite connected and automated vehicle (CAV) technology revolutionarily enables individual-level cooperative control, most existing studies have limited capability in simultaneously considering lane-changing and car-following behaviours of vehicles while passing through continuous intersections. This study proposes a novel trajectory optimisation method for a mixed fleet of CAVs and human-driven vehicles, including lateral and longitudinal behaviour control. The method constructs a lane-changing trajectory optimisation model based on the lane-changing intention generated by CAVs and formulates safety constraints, lane occupancy state assignment and lane-changing cost function constraints. It also establishes a longitudinal following model based on vehicle role switching protocols to realise different constituent fleets passing through signalised intersections with minimal stops. Simulation results show that the proposed method can reduce the average travel time by up to 26.77%, decrease average travel delay by up to 42.66%, minimise the average number of stops by up to 91% and lower fuel consumption and pollutant emissions by more than 28%. After multiple independent experiments are conducted, the 95% confidence intervals are determined as follows: [101.09 s, 101.53 s] for average travel time, [49.77 s, 50.17 s] for travel delay and [111.22 g, 132.38 g] for fuel consumption. Parking behaviour analysis yields [0.68 s, 0.86 s] for average parking time and [0.16, 0.18] occurrences per vehicle for stop frequency. Sensitivity analyses of signal cycle length and guidance zone length demonstrate that the guidance effect of the proposed control strategy is stable under various settings.