Adaptive Headlight Simulation

In this project, I worked in a team of 5 students, and we used a webcam-based perception module (OpenCV Simple Blob Detector and LAB color space conversion) for image processing on incoming vehicles. It enabled the dynamic dimming or brightening of individual LED segments based on vehicle positioning. It also provides a distance estimation in approaching vehicles' x and z coordinates for precise control. Additionally, there is also a GUI that provides real-time visualization of the adaptive headlight grid. We tested this system using CARLA in a realistic driving environment to evaluate performance based on conditions. We were able to achieve a detection accuracy of 92% and a 65% reduction in glare.