Autonomous Car Algorithm Development

Dates: May 2024

Class: ECE484 - Final Project

This project involved developing an autonomous racing algorithm using python for the GRAIC competition using ROS and CARLA. The focus was on advanced computer vision, intelligent pathfinding, and obstacle avoidance systems, achieving superior lap times and dynamic responsiveness.

Project Highlights

- Pathfinding: Implemented RRT with gradient descent for optimal racing lines.
- Control Systems: Enhanced PID control for smooth braking and momentum conservation.
- Computer Vision: Real-time obstacle detection and predictive avoidance strategies.

Performance Results

The algorithm consistently outperformed benchmarks, maintaining high cornering speeds while effectively avoiding obstacles. The optimization of racing lines through advanced algorithms significantly improved overall performance and responsiveness.

This embedded video shows one of our final runs with the RRT pathfinding: