The U.S. Army Engineer Research and Development Center (ERDC) Virtual Autonomous Navigation Environment (VANE) development team recently developed new methods for determining how to effectively deploy sensors using optimization techniques guided by high-fidelity simulations. The group was able to simulate a vehicle-mounted infrared sensor for analysis purposes, resulting in successful detection of disturbed soil and additional benefits that ultimately contribute to the safety of the Warfighter. The team was awarded the 2013 Army Modeling and Simulation (M&S) Award in the analysis category for this development.
“In the analysis, the sensor’s ability to detect the region of disturbed soil was predicted for several lens properties and mount positions on the vehicle, and the optimum configurations were determined by finding the maximum probability of detection for the disturbed soil region,” said Geotechnical and Structures Laboratory (GSL)’s Dr. Chris Goodin, VANE team member.
The ERDC-developed VANE is high-fidelity, physics-based software that simulates terrain and environmental conditions and the ways they impact sensor performance. The technology was initially intended for use in simulating robotic vehicles but transitioned following the development of an M&S capability for simulating radiometric properties of GPS, LiDAR, hyperspectral imagers and thermal infrared sensors. Those sensors were then applied to realistic and complex terrain representations using ERDC’s Geo-Environmental Tactical Sensor Simulation software, enabling analysis to predict the ideal standoff distance for vehicle-mounted infrared sensors and optimal route analysis for GPS-enabled vehicles.
“The VANE simulation capability can be used to conduct a wide range of analyses,” said Goodin. “For example, it was used during a recent collaboration with the Army Material Systems Analysis Activity to develop routing algorithms for navigating urban areas with GPS-based perception systems. We were able to predict where GPS signal dropout was most likely to occur in a typical urban environment and these areas of high-sensor error were encoded into an obstacle map, which was then used to plan a route around these potentially hazardous areas.”
The technology currently supports ERDC’s Vehicle Mounted Perceptions Systems program, which provides physics-based M&S tools and data to Department of Defense agencies involved in ground-vehicle sensor systems development and testing. In the coming years, the VANE team, which includes researchers from GSL, as well as ERDC’s Information Technology and Environmental laboratories, plans to improve and develop additional physics models. They also hope to expand into a wider range of analyses. The team is also collaborating with the Tank and Automotive Research, Development, and Engineering Center to simulate the performance of sensor systems used in autonomous navigation.
“By using the VANE’s detailed environmental simulation, the performance of the autonomous system can be predicted for types of environments in which the robot has never been tested," said Goodin. "This capability will allow future software systems that need to be 'trained' to be quickly deployed into new environments. As increasing numbers of ground vehicles begin to rely on sensor-based perception systems to keep drivers safe and effective, VANE will be used to ensure these systems are reliable, optimized and ready for a wide-range of operational environments.”