Real-time vision-based obstacle detection & avoidance for laptop and Raspberry Pi
It solves the problem of detecting and avoiding obstacles using a single webcam, providing real‑time object detection, tracking, Kalman filtering, RK4 trajectory prediction and collision risk assessment. The system runs a multi‑threaded pipeline that keeps the reactive loop at full FPS, and can optionally add high‑level scene reasoning via OpenAI without blocking. It targets developers and hobbyists building robotics or safety‑critical applications on laptops or Raspberry Pi devices. Compared to generic detection scripts, it integrates physics‑based prediction and planning, offering a more reliable and extensible solution.
View on GitHub →gurnoorssandhu/vision_modal