This paper proposes a game theory-based lane changing decision-making approach for the highway diversion scenario, which is applicable to the mixed traffic environment composed of autonomous vehicles and human-driven vehicles. The lane changing decision-making process is modeled as a non-cooperative two-player game, a dynamic game mechanism under incomplete information is established considering driver heterogeneity and decision diversity, and a complete strategy set in normal form is derived using the Harsanyi transformation. The parameters of the decision model are estimated based on the NGSIM dataset and a bi-level calibration framework, with the upper-level using stochastic gradient descent to reduce the estimation error and the lower-level solving the Perfect Bayesian equilibrium. The true positive rate and detection rate are used as indicators to verify the high behavioral prediction accuracy of the proposed approach. Simulation experiments are carried out to validate that the proposed model has good efficiency and safety performance. The results indicate that for different traffic flow densities, the proposed model has higher lane changing efficiency compared to the selected baseline and has stable reciprocal of time to collision.
A Non-Cooperative Dynamic Game-Based Approach for Lane Changing Decision-Making in Mixed Traffic Diversion Scenarios
2023-09-24
1120924 byte
Conference paper
Electronic Resource
English
Cooperative Decision-Making in Mixed Urban Traffic Scenarios
DataCite | 2023
|Case-based decision making for diversion scenarios
IEEE | 2023
|Automatic driving forced lane changing decision-making method based on dynamic continuous game
European Patent Office | 2024
|European Patent Office | 2024
|European Patent Office | 2020
|