Quantifying and Improving Laser Range Data When Scanning Industrial Materials

摘要

This paper presents the procedure and results of a performance study of a miniature laser range scanner, along with a novel error correction calibration. Critically, this paper investigates the accuracy and performance of the ranger sensor when scanning large industrial materials over a range of distances. In addition, this paper investigated the effects of small orientation angle changes of the scanner, in a similar manner to which it would experience when being deployed on a mobile robotic platform. A detailed process of error measurement and visualization was undertaken on a number of parameters, not limited to traditional range data but also received intensity and amplifier gain. This paper highlights that significant range distance errors are introduced when optically laser scanning common industrial materials, such as aluminum and stainless steel. The specular reflective nature of some materials results in large deviation in range data from the true value, with mean root mean square error (RMSE) errors as high as 100.12 mm recorded. The correction algorithm was shown to reduce the RMSE error associated with range estimation on a planar aluminum surface from 6.48% to 1.39% of the true distance range.