Game-Theory-Lane-Changing

• Implement two-player game theory lane chaning with python, and visulaize with MATLAB Automated Driving Toolbox.
• The predinfined routes are extracted from NGSIM dataset, and refer to Multi-Player Dynamic Game-Based Automatic Lane-Changing Decision Model under Mixed Autonomous Vehicle and Human-Driven Vehicle Environment.
• The game theory model and payoff functions refer to Modeling Lane-Changing Behavior in a Connected Environment: A Game Theory Approach.

Usage

1. Genrate game theoretic simulation trajectory file.

# 'trajectory_post.csv' contains predefined routes of vehicles, and are extracted from NGSIM dataset
# use simulation.py to simulate the trajectory of vehicles 
# where the target vehicle interacted with the lag vehicle with the predifined two-player game when performing lane channging
# output 'trajectory_A.csv' for further visualization using MATLAB
python simulation.py

2. Visualize the trajectory file.

compile and run simulation_traj.m

3. Analyze the lane changing trajectory

compile and run plot_3d.m, plot_traj.m

Simulation Result

1. In DrivingScenario directory, there are driving scenarios files that can be visualized by MATLAB automated driving toolbox.

• rule_based.mat: Vehicle A based on rule-based lane changing, and fails because of safety distance
• data_based.mat: The visualization of real world data from NGSIM data, vehicle A fails to change lane.
• game_theory.mat: Vehicle A succeed to change lane via game with Vehicle B, while considering safety distance.

2. In video directory, there are videos of the visualization of the three driving scenario files.