Table 2.

The table shows the root mean square errors of the cameras and LiDARs before and after the Bundle Adjustment (BA)-like technique defined in Section 5. The errors are measured in pixels and meters for the cameras and LiDARs, respectively. The synthetic testing scene consist of two cameras and two LiDARs. The real-world test is done with a Velodyne HDL-64, two Velodyne VLP-16 and two cameras. Additionally Gaussian noise is added to the synthetic LiDAR point cloud with $0.04$ standard deviation.

	Before BA		After BA
	Camera	LiDAR	Camera	LiDAR
Synthetic	4.231 px	0.02231 m	2.152 px	0.02001 m
Real-world	2.892 px	0.03515 m	0.963 px	0.01036 m