← Back to projects
FEB 2026 – MARCH 2026
Autonomous Racecar Perception & Control
Python, C++, ROS 2, OpenCV
- Built a camera-based visual servoing pipeline (HSV segmentation, SIFT, YOLO, homography ground-plane projection) for real-time detection and localization, reaching 0.79 median IoU and 1.5 cm mean positioning error
- Designed a LiDAR wall-following controller (PD control + RANSAC line fitting) and an independent >180°-FOV safety controller, achieving stable 0.55–0.71 m wall tracking at 1–2 m/s with sub-0.25 m emergency braking
- Implemented and benchmarked A* and RRT planners over an occupancy grid and a pure-pursuit controller with off-path recovery, completing the 400 m race lane at a 95% success rate on a physical autonomous racecar