# Design and Error Calibration of a Machine Vision-Based Laser 2D Tracking System

## TL;DR

This paper introduces a high-precision laser tracking system using machine vision and an off-axis optical design to improve accuracy and reduce errors.

## Contribution

The novel contribution is a machine vision-based laser tracking system with an off-axis design and SVD-based error calibration for improved precision.

## Key findings

- The system achieves a spatial RMSE of 0.189 mm at 1.5 m working distance.
- The off-axis design reduces dependency on optical coaxiality and mechanical precision.
- SVD-based calibration improves tracking accuracy and system reliability.

## Abstract

A laser tracker is an essential tool in the field of precise geometric measurement. Its fundamental operating idea is a dual-axis rotating device that propels the laser beam to continuously align and measure the attitude of a collaborating target. Such systems provide numerous benefits, including a broad measuring range, high precision, outstanding real-time performance, and ease of use. To solve the issue of low beam recovery efficiency in typical laser trackers, this research offers a two-dimensional laser tracking system that incorporates a machine vision module. The system uses a unique off-axis optical design in which the distance measuring and laser tracking paths are independent, decreasing the system’s dependency on optical coaxiality and mechanical processing precision. A tracking head error calibration method based on singular value decomposition (SVD) is introduced, using optical axis point cloud data obtained from experiments on various components for geometric fitting. A complete prototype system was constructed and subjected to accuracy testing. Experimental results show that the proposed system achieves a relative positioning accuracy of less than 0.2 mm (spatial root mean square error (RMSE) = 0.189 mm) at the maximum working distance of 1.5 m, providing an effective solution for the design of high-precision laser tracking systems.