2013 IR&D Annual Report

Investigation into Techniques for Detecting Negative Obstacles 10-R8278

Principal Investigators
Steven W. Dellenback
Jason Gassaway
Richard Garcia

Inclusive Dates:  01/01/12 – 01/12/12

Background — The U.S. military has repeatedly stated the single most complex, unresolved issue in the Unmanned Ground Vehicle (UGV) community is the detection of negative obstacles (these are holes, troughs and anything that does not reflect or is difficult to observe). In most cases, it is simply a geometry problem because the angle in which sensors can detect a hole is not acute enough until the vehicle is too close to do anything about it.

Approach — The goal of this project was to develop techniques that facilitate the detection of “negative obstacles” for UGVs. Negative obstacles are very difficult to identify from a traditional UGV platform moving at a significant speed because the elevation of the sensors does not provide a field-of-view (FOV) that can identify the obstacle in enough time to perform avoidance maneuvers. Techniques currently used are to place downward looking sensors at the front of the vehicle and have a FOV that is measured in a small number of meters with a UGV moving at a very slow rate of speed. This effort investigated the use of sensors that will be located above the vehicle using an airborne platform.

Accomplishments — The following activities were completed:

  • Designed and fabricated a mast structure to allow the sensors to be located 12 meters above a HMMWV.
  • Designed an algorithm framework that implements a pipeline processing framework composed of the following steps: data collection, localization, filtering, analysis and detection.
  • Implemented algorithms to capture data from both LIDAR and a camera attached to the mast. The sensors capture scans of the negative obstacle test area while mounted to the mast and the results provide highly reliable identification of negative obstacles. The following figure depicts how the detection efforts were implemented.
SwRI researchers developed a set of algorithms for use by an airborne sensor that uses LIDAR and vision data to detect holes and depressions not normally detectable by vehicle sensors.
SwRI researchers developed a set of algorithms for use by an airborne sensor that uses LIDAR and vision data to detect holes and depressions not normally detectable by vehicle sensors.
Benefiting government, industry and the public through innovative science and technology
Southwest Research Institute® (SwRI®), headquartered in San Antonio, Texas, is a multidisciplinary, independent, nonprofit, applied engineering and physical sciences research and development organization with 10 technical divisions.
04/15/14