autonomous_racing

• ROS package for autonomous racing using optimal control
• Contains ROS nodes to run high level planner and low level tracking controller together

Description:

High level:

• pdgplanner.py is the high-level planner class of the Ego vehicle
• aiagent.py contains competing AI agent class
• high_level_node is the ROS node calling these classes

Low Level:

• tracking_MPC_allConstr.mexa64 is the the controller compiled from ACADO Toolkit
• nlmpc_autorace_sim_allConstr.m is the low level control simulation ros node

Key dependencies:

• pdgplanner.py should be in the same directory as ilqr/ folder. ilqr code is used from https://github.com/anassinator/ilqr
• aiagent.py requires Casadi in the appended path
• ROS
• MATLAB
• Python 2.7
• Numpy
• Theanos
• ACADO toolkit for MATLAB (https://acado.github.io/matlab_overview.html)
• ROS Toolbox from MATLAB

Instructions to run:

$ cd ~/catkin_ws/src
$ git clone https://github.com/pgupta2050/autonomous_racing.git
$ roscore
$ rosrun autonomous_racing high_level_node.py

Then run the MATLAB file nlmpc_autorace_sim_allConstr

Simulation result for a circular track with 1 competing agent

Time snaps: ego in blue and competitor in red

Contributors:

• Viranjan Bhattcharyya
• Jacky Tang
• Prakhar Gupta