Bos Group on Testing of Lidar for Autonomous Vehicles

Colorful lidar image of an outdoor area in one image, with a near vertical green line in the second image.
(a) The reference point cloud scan (gray) overlayed with point clouds collected by each of the DUT lidars (colors). (b) Side view of an initial alignment between the reference point cloud (green) and point clouds from the DUT lidars for the 10 m target. Notice that the target is tilted toward the test origin. See the open source article link below.

Jeremy Bos (ECE) was quoted and PhD student Zach Jeffries (electrical engineering) and Akhil Kurup ’22 (PhD, computer engineering) were mentioned by SPIEGreen Car CongressTech and SCIENMAG in a story about a three-year effort to develop tests and performance standards for lidars used in autonomous vehicles and advanced driver assistance systems.

Bos led the testing through its first year, with Jeffries’ assistance. The team’s findings are detailed in an open-access paper published this month in Optical Engineering.

Zach D. JeffriesJeremy P. BosPaul F. McManamon, Charles Kershner, Akhil M. Kurup
Optical Engineering, Vol. 62, Issue 3, 031211 (January 2023).


This paper describes the initial results from the first of 3 years of planned testing aimed at developing methods, metrics, and targets necessary to develop standardized tests for these instruments. Here, we evaluate range error accuracy and precision for eight automotive grade lidars; a survey grade lidar is used as a reference. These lidars are tasked with detecting a static, child-sized, target at ranges between 5 and 200 m.

Our purpose in this work is to motivate the development of test standards in this area and highlight variations in performance between lidars when stated specifications are similar.

Proposed additions to the testing include more complex targets, dynamic targets, placing corner cubes, or identical lidars on the test range, and weather effects.