4_6

Iterative Closest Point (ICP)

• Load a lidar point cloud datum from KITTI dataset

1. Based on known correspondences, use ICP to find the transformation between two point clouds
2. Without known correspondences, use ICP from PCL library to find the transformation between two point clouds

In both cases, the point clouds are rotated by 10 degrees around the z-axis, and moved 2m forward.

---

How to build & run

Requirement: PCL

Local build

mkdir build && cd build
cmake ..
make -j
cd ..
./build/known_corr_icp
./build/pcl_icp

Docker build

Requires base build

docker build . -t slam:4_6
docker run -it --env DISPLAY=$DISPLAY -v /tmp/.X11-unix/:/tmp/.X11-unix:ro slam:4_6

# Inside docker container
cd fastcampus_slam_codes/4_6
./build/known_corr_icp
./build/pcl_icp

---

Output

PCL_ICP

Green - Source point cloud Red - Target point cloud Blue - Transformed source point cloud

Purple is overlap of various colors