How jayaschool1981-rgb/ai-based-autonomous-navigation-system works
Project type and purpose
• This is an embedded robotics system rather than a web or desktop application.
• Purpose: enable a small vehicle (Raspberry Pi/Jetson-Nano-powered) to navigate autonomously by fusing vision (YOLO-tiny object detection, OpenCV lane detection), ultrasonic sensing, and A* path planning, driving DC motors via an L298N controller.
File structure
• models/
• coco.names, yolov3-tiny.cfg → YOLO model definition and classes for object detection
• requirements.txt → Python dependencies (OpenCV, PyTorch/TorchVision or Darknet bindings, NumPy, etc.)
• src/
• camera_capture.py → captures frames from USB/Pi camera
• yolo_detector.py → runs YOLO-tiny on each frame
• lane_detection.py → runs Hough-transform-based lane extraction
• ultrasonic_sensor.py → reads distance from HC-SR04
• astar_planner.py → computes collision-free route using A*
• motor_controller.py → drives L298N PWM signals for differential steering
• main_robot.py → orchestrator: calls capture → detect/lane/sense → plan → drive
README for additional insights
• Hardware stack: Pi/Jetson, camera, ultrasonic, L298N + motors
• Key features list confirms perception, planning, control pipeline
• Simulation mode hint: code can run without hardware—add a “Simulated Sensor/Actuator” sub-component when drawing
Be very detailed—capture every architectural element
• Don’t hesitate to show sub-components (e.g., split computer vision into “Object Detector” and “Lane Detector”)
• Both real-world hardware and software abstractions