An autonomous vehicle prototype capable of navigating a colored track using computer vision. It processes camera input in real-time to adjust steering and throttle.
Navigating a physical track with varying lighting conditions using only a single low-cost camera.
Used OpenCV for lane detection (color thresholding + edge detection). A basic regression model predicts the steering angle based on lane curvature. The Python controller runs an optimized high-frequency loop on the Pi.
Embedded System Architecture. The Raspberry Pi runs a Python main loop that polls the Camera, processes the frame (OpenCV), calculates error (PID Controller), and adjusts PWM output to the Motor Driver HAT.
Calibrate camera and sensors.
Capture frame and preprocess.
Calculate deviation from center line.
Adjust servo and DC motor duty cycle.