F1-TenthSep 01 2023 - May 6 2024
F1/10th is an educational and research platform that uses 1/10 scale autonomous race cars to develop and test autonomous driving technology. Participants build small cars equipped with sensors like LiDAR and cameras, using ROS software to implement functions such as path planning and obstacle avoidance. The project serves both as a tool for learning robotics and for researching complex autonomous driving algorithms, with competitions providing hands-on practice and testing.
What we did
Model
Predictive control (MPC) is an optimal control technique that the calculated control
actions minimize a cost function for a constrained dynamic system over a
finite, receding, horizon.
For
F1 tenth autonomous driving vehicle, simple PID control, pure pursuit and RRT
for avoiding obstacles can allow the car to complete the path following, but
the application of MPC technology can make the car's path following more
efficient and improve the speed and safety of the car. The goal of MPC is to
generate valid input controls for T steps ahead that controls the vehicle and
follow the reference trajectory as close as possible.
Our reposotory: https://github.com/qh65/F1-Tenth-Duke
SLAM LiDAR scan test VESC test
SlAM map scan and generation Particle filter in Gazebo sim