Vision pipeline for autonomous ground vehicle

Course: Probabilistic Mobile Robotics, Indian Institute of Technology Kanpur, 2017

Developed vision based systems for autonomous maneuvering of an intelligent ground vehicle in traversing within lanes, avoiding obstacles and maintaining its heading while navigating in unfamiliar territory.

  • Designed an algorithm for robustly detecting lane regions in a video feed using PCA with region proposals from previous frames.
  • Combined LiDAR and camera inputs to seperate obstacles and lanes in an image, as well as generate a semantic map for motion planning algorithms.

Report Presentation