Sensor Fusion Self-Driving Car Course, Lidar

Objects detection from scratch

In this repo, I show how to do pointcloud segmentation and clustering from scrath. Though there are functions in PCL library to do this task more efiiciently I found it pretty beneficial to also learn a working fundamental.

Implemenation includes:

  • RANSAC to separate between plane and obstacles.
  • KD-tree insert and search to help speed up clustering process.
  • Euclidence clustering.

Other processed that get some help from PCL library:

  • Downsampling using voxel grid.
  • Bounding box.
  • Visualization.