lidar_lidar_calib

LiDAR-to-LiDAR extrinsic calibration using GICP (Generalized ICP) point cloud registration. Computes the rigid transform between the center LiDAR and each side LiDAR (NE, NW).

Overview

The NE and NW LiDARs are side-mounted and their transforms relative to the center LiDAR must be known for lidar_aggregator to merge the point clouds correctly. This package takes simultaneous scans from two LiDARs and finds the transform that best aligns them.

GICP is an iterative refinement algorithm — it requires a reasonable initial guess (from physical measurement) to converge. The calibration must be run with the vehicle stationary in a feature-rich environment.

Prerequisites

• Vehicle must be stationary
• Environment must have sufficient geometric features (not an empty, symmetric room)
• A physical measurement of the approximate LiDAR mount positions must be available as the initial guess

Usage

colcon build --packages-select lidar_lidar_calib
source install/setup.bash
ros2 launch lidar_lidar_calib lidar_lidar_calib.yaml

This launches two calibration instances in parallel — one for NE and one for NW — each registering its side LiDAR against the center LiDAR. The resulting transforms are logged and published.

Once calibration converges consistently, update the corresponding <origin> blocks in eve_description/urdf/ with the calibrated values.