Calibrating several LiDARs to each other, by hand or automatically
A machine with several LiDARs needs each sensor placed in one common coordinate system. Livox describes a manual method and an open-source automatic method that needs no reference object.
At a glance
Results in figures
- Manual method accuracy
- Within 0.1°
- Source: Livox
The challenge
Turning several sensors into one coordinate system needs the position and angle of each one (extrinsic calibration). Doing it by hand needs a shared field of view and a suitable reference object, which is not always available.
What was done
- Manual: in Livox Viewer, adjust six values (x, y, z, roll, pitch, yaw) until the point clouds of a reference object overlap.
- Automatic (TFAC-Livox, Target-Free Automatic Calibration): one LiDAR builds a map by SLAM. The second LiDAR's points are moved using timestamps, the motion path and a rough starting estimate, then matched to the map until the calibration settles.
- Livox advises using a feature-rich place such as an underground car park, avoiding moving objects and moving slowly, especially at corners. The LiDARs do not need overlapping fields of view for the automatic method, as long as starting values are known.
Results
- Livox gives one figure: the manual method can be accurate to within 0.1 degree.
- It gives no figures for the automatic method. The code is published on GitHub as Livox_automatic_calibration.
Equipment
Livox Viewer and two Livox Horizon LiDARs. The Horizon is an earlier Livox model; for current sensors see our Livox LiDAR range.
About this case study
This showcase was published by Livox on 23 June 2020; Livox products are among those we supply in India. Ecoleap Solutions wrote this summary; figures are as reported by the manufacturer.
Planning a survey, a tender or a new project?
Send us your scope and we will reply with a suitable configuration and a budgetary quote, usually within one working day.