TSwR Projekt - MPC, iLQR, Stanley, Pure Pursuit

Description

A project implementing four controllers for a vehicle in the AWSIM environment using ROS2 Humble.

Installation Guide

1. Enter the autoware container -> run_x86.sh
2. cd ~/autoware/src
3. git clone https://github.com/mgmike1011/tswr_awsim
4. cd ..
5. colcon build --packages-select tswr_awsim
6. . install/setup.bash

Controllers

Run all comands inside docker container

iLQR

Installing additional required packages

pip install drake

1. Start AWSIM simulator

foo@bar:~/autoware$ cd f1tenth_simulators/F1Tenth_v0.5

foo@bar:~/autoware/f1tenth_simulators/F1Tenth_v0.5$ ./F1Tenth_v0.5.x86_64 

2. Start path publisher

foo@bar:~/autoware$ ros2 run tswr_awsim path_publisher

3. Start iLQR planning algorithm

foo@bar:~/autoware$ ros2 run tswr_awsim iLQR_controller

Linearized Model Predictive Control:

Computing steering angle and acceleration (with some limitations) based on current pose and a reference trajectory. The controller uses linearized model of vehicle kinematics to predict future state. It solves a discrete-time algebraic Riccati equation (DARE) to obtain the optimal feedback gain matrix K, which is used to compute optimal control input.

1. Start AWSIM simulator

foo@bar:~/autoware$ cd f1tenth_simulators/F1Tenth_v0.5

foo@bar:~/autoware/f1tenth_simulators/F1Tenth_v0.5$ ./F1Tenth_v0.5.x86_64 

2. Start Linearized Model Predictive Control algorithm:

foo@bar:~/autoware$ ros2 launch tswr_awsim lin_MPC_launch.py

Pure Pursuit:

Computing steering angle using vehicle kinematics (with wheelbase 0.33 m), reference path and lookahaead distance 0.7 m. Equations correspond to a vehicle model with reference point in the center of the rear axle.

1. Start AWSIM simulator

foo@bar:~/autoware/f1tenth_simulators/F1Tenth_v0.5$ ./F1Tenth_v0.5.x86_64 

2. Start Pure Pursuit algorithm:

foo@bar:~/autoware$ ros2 launch tswr_awsim pure_pursuit_launch.py 

Stanley

Computing steering angle using the heading error and cross-track error. In this method, the cross-track error is defined as the distance between the closest point on the path with the front axle of the vehicle.

1. Start AWSIM simulator

foo@bar:~/autoware/f1tenth_simulators/F1Tenth_v0.5$ ./F1Tenth_v0.5.x86_64 

2. Start Stanley algorithm:

foo@bar:~/autoware$ ros2 launch tswr_awsim stanley_launch.py 

Additional literature

• iLQR
• iLQR
• Pure Pursuit, Stanley
• Stanley

Authors

• Agnieszka Piórkowska
• Miłosz Gajewski
• Maciej Mirecki
• Mikołaj Zieliński

Politechnika Poznańska 2023