jeffrey28/auto-driving-tsinghua

Bai Liu's bachelor thesis project, on optimizing control strategies of auto-driving vehicles at crossroads

Intersection Routing Strategy for Self-driving Vehicles

Welcome! This project is created by Bai Liu.

How to manipulate?

1. Open MATLAB
2. Locate "Current Folder" of MATLAB to where the source codes are
3. Run aimed part(s) by editting "Main.m" (adding or removing relevant comments)
4. Adjust parameter(s) in "InitializeGlobal.m"
5. Run "Main.m"

Structure of the system

The project has three optimization objects: left-turning strategy in single-agent case, left-turning strategy in multi-agent case, and traffic signal display strategy. Besides, several files serve as supporting tools.

1. Console

   "Main.m": core console, the control center

   "InitializeGlobal.m": define and initialize global variables, including parameters of vehicles, road structure and simulation method

2. Optimization on left-turning strategy in single-agent case

   We adopt Q-Learning method here.

   "OptSingleTurning.m": algorithmic implementation of Q-learning

   "CalAction.m": calculates possible future states for a given state in Q-learning, to assist "OptTurning.m"

   "TestSingleTurning.m": helps to illustrate the training result

3. Optimization on left-turning strategy in multi-agent case

   To be implemented.

4. Optimization on traffic signal display strategy

   We adopt SA (simulated annealing), GA (genetic algorithm) and PSO (particle swarm optimization) here.

   "OptSignal.m": Implement SA, GA and PSO

5. Investigation on how multiple factors impact the optimization effect

   To be implemented.

6. Miscellaneous

   "XroadSimulation.m": simulate the traffic conditions of the intersection for a given amount of time

   "CalVehicleNum.m": calculate the number of vehicle(s) to be geneerated in a given amount of time

   "GenerateVehicle.m": generate vehicle objects