Next: Stereo Vision for Up: Sensor and Processor Previous: Task Space Estimation

Range Measurement Systems

Authors: [tex2html_wrap4214]G. Soucy, T. Baron, P. MacKenzie, P. Tremblay, P. Whaite, F.P. Ferrie

Investigator username: ferrie

Category: perception

Subcategory: sensor and processor design

Initial work in range measurement systems at CIM was prompted by a joint project with the Institute for Information Technology at the National Research Council of Canada in 1986. The goal of that project was to design a compact laser rangefinding system for research and applications in robot vision based on the Rioux synchronized scanning technology. A compact scanner was developed, weighing less than [tex2html_wrap4206] and with a field of view of approximately [tex2html_wrap4208]. It can produce a [tex2html_wrap4210] range image at 10 bits/pixel, with an accuracy of [tex2html_wrap4212] at closest approach. It has been an important tool in our research and has proved invaluable since it began operation in early 1990.

Development work still continues on the scanner and support software. Algorithms have been developed for adapting laser power to account for changes in surface reflectance, as well for improving calibration accuracy. An interactive tool called RIM was also developed to provide a convenient environment for data acquisition, visualization, and control over camera parameters such as the programmable field of view. Although the design of the scanner is relatively stable, further improvements are anticipated as new technologies become available.

The extension of our research into the areas of mobile robotics and CAD/CAM systems has prompted the development of other range measurement systems based on light striping and active focusing (BIRIS). Although each of these is specialized to a particular application domain, system design has focused on the development of common tools for data acquisition, calibration and interpretation. This has lead to the a series of virtual instruments that are transparent to the end user, differing only in field of view, quantization and accuracy.


baron@cim.mcgill.ca