Speed Planning Strategy in Urban Environments Considering Passenger’s Preferences, Traffic Signals and Road Limits

A personalized speed planning strategy for autonomous vehicles (AVs) is proposed in this paper, which can generate individual speed trajectories for passengers with different driving preferences in typical urban environments with multiple traffic signals, road terrain, and speed limits. Firstly, the K-means method is used to analyze and cluster the drivers’ data in the case of free driving (not following the front vehicle) in the NGSIM dataset, and then the actual longitudinal speeds and accelerations preference of different drivers’ groups in the urban environments can be obtained. Secondly, according to the information on road restrictions, traffic signals phase, and driving preferences of the passengers in AVs, a decision-making mechanism is designed to calculate the personalized traveling time of each traffic signal section, in which the preferences of speeds are transformed into the preferences of traveling time. Then, the dynamic programming (DP) algorithm and the bisection method are integrated to plan personalized speed trajectories for each traffic signal road section, with consideration of the factors such as preferred traveling time in this road section, road speed limits, passenger comfort, driving economy, etc. Finally, a typical urban scenario is designed for algorithm verification and performance evaluation. The test results show that the strategy can generate safe, comfortable, energy-saving, and personalized speed trajectories for different groups of passengers.