Comparing the performance of structured light depth sensors and traditional time-of-flight depth sensors for use in a lunar mining environment

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
University of Alabama Libraries

Autonomous robots are seen as a necessary component for long term manned missions to the Moon. The robots are necessary to excavate lunar soil for processing for in situ resource utilization. The lunar environment poses several challenges to autonomous robotic navigation and the choice of sensor technologies is more restricted than on Earth. Without GPS and ultrasonic technologies, localization and obstacle detection are often performed using data from a laser-based scanner. Laser scanners have been used in robotics on Earth for many years to provide the distances to surrounding objects. Newer sensors, based upon the use of structured light, can provide range data faster and at a lower cost than traditional laser scanners. The purpose of this project is to evaluate a structured light depth sensor, the Microsoft Kinect for Xbox 360, and a traditional multi-echo laser scanner, the Hokuyo UTM-30LX-EW, to determine if they are suitable for autonomous robotic navigation tasks in a lunar mining application. Experimental results are presented that indicate that IR saturation will prevent the Kinect from producing usable distance data. While IR does not affect the lidar, suspended dust in the environment adversely affect both sensors, differently. In dusty environments, the Kinect performs better at shorter distances while the lidar performs better at longer distances to target. The results indicate that a hybrid system utilizing a Kinect for short range obstacle detection and avoidance combined with a lidar for long range landmark identification and localization could serve as a solution in dusty lunar mining environments protected from excessive IR saturation.

Electronic Thesis or Dissertation
Computer engineering, Robotics