Advanced Lane Finding

Autonomous Vehicle Engg

The Project

The goals and steps of this project are:

• Compute the camera calibration matrix and distortion coefficients from a set of chessboard images.
• Apply distortion correction to raw images.
• Apply a perspective transform to generate a bird's-eye view.
• Use color transforms and gradient thresholds to create a binary image.
• Detect lane pixels and fit polynomial curves to find the lane boundaries.
• Determine the lane curvature and vehicle position with respect to the lane center.
• Warp the detected lane boundaries back onto the original image.
• Produce an output visualization with lane boundaries and numerical estimations for curvature and vehicle offset.

Data

• Calibration images: stored in the camera_cal/ directory.
• Test images: stored in the test_images/ directory for pipeline testing on single frames.
• Videos:
  • project_video.mp4 – Main test video.
  • challenge_video.mp4 – Optional, moderate difficulty.
  • harder_challenge.mp4 – Optional, much harder test.

If you're feeling ambitious, try recording and testing your own driving videos! Calibrate your camera and implement the pipeline from scratch.

---

Usage

1. Set up the environment

conda env create -f environment.yml