Lane changes can lead to disturbances in traffic flow, whilst the uneven distribution of traffic over different lanes as a result of lane changes can also lead to instabilities and congestion on one specific lane. Therefore, giving advice on lane change can be beneficial for both individual drivers and traffic state in the network. However, there are many variations in advice content and objective, all of which may impact the performance of advice. This paper focuses on the optimization of traffic flow through the performance of specific lane changes. The authors model traffic flow on a two-lane stretch and consider lane-change time instants of a subset of vehicles as decision variables. Optimizations with three objectives are constructed: reaching a collective optimum, a group optimum and a user optimum. These optima are found by total travel delay minimization of different vehicle groups. To solve the problems, a genetic algorithm as a heuristic method is implemented. Each optimum leads to different lane changes. Specifically, by the proposed algorithm, vehicles will be suggested to change lanes in bigger gaps to improve collective or group efficiency; while they are supposed to overtake as many vehicles as they can by changing lanes for their own benefit. The algorithm can be further extended to a more effective in-car advice system, which can improve traffic efficiency for future situations through communicating partly automated vehicles.@en

Lane changes can lead to disturbances in traffic flow, whilst the uneven distribution of traffic over different lanes as a result of lane changes can also lead to instabilities and congestion on one specific lane. Therefore, giving advice on lane change can be beneficial for both individual drivers and traffic state in the network. However, there are many variations in advice content and objective, all of which may impact the performance of advice. This paper focuses on the optimization of traffic flow through the performance of specific lane changes. The authors model traffic flow on a two-lane stretch and consider lane-change time instants of a subset of vehicles as decision variables. Optimizations with three objectives are constructed: reaching a collective optimum, a group optimum and a user optimum. These optima are found by total travel delay minimization of different vehicle groups. To solve the problems, a genetic algorithm as a heuristic method is implemented. Each optimum leads to different lane changes. Specifically, by the proposed algorithm, vehicles will be suggested to change lanes in bigger gaps to improve collective or group efficiency; while they are supposed to overtake as many vehicles as they can by changing lanes for their own benefit. The algorithm can be further extended to a more effective in-car advice system, which can improve traffic efficiency for future situations through communicating partly automated vehicles.@en