Lidar design, use, and calibration concepts for correct environmental detection

摘要

The useful environmental interaction of a mobile robot, is completely dependent on the reliable extraction of information from its immediate surroundings. A particular class of sensors often now applied to this problem is the lidar system. The aim of the article is to examine the performance limits and sources of error in these sensors at their design and calibration stages and during their general use. A framework, aimed directly at optimizing the quality of the output information, for use in mobile robot navigational algorithms, is given. The design concepts for producing correct range estimates in the presence of a large dynamic range of surface albedo is addressed. The performance limits, which can be expected in terms of systematic and random range errors, are theoretically analyzed and modeled to provide a correct calibration procedure. During this derivation, it is shown that the naive determination of the sensor to target distance as a function of any lidar's output signal, in general provides a false calibration. The possible scanning speed and data sampling rates are derived as functions of a lidar's geometrical and electronic temporal design specifications. Finally the issue of temporally averaging of several range values is demonstrated and it is shown that under certain quantified conditions, range variance reduction is possible. The text addresses the use of amplitude modulated continuous wave (AMCW) and time of flight lidars in general, but makes several references to a particular lidar design example, giving results and conclusions from an actual engineering AMCW lidar implementation.