System and method for comparing simulated environments for unmanned ground vehicle development and testing
Inventors
AUCHTER, Joseph A. • Towler, Jerry A. • BROTHERS, Robert J.
Assignees
Southwest Research Institute SwRI
Publication Number
US-12287633-B2
Publication Date
2025-04-29
Expiration Date
2042-09-26
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
In an approach to comparing simulated environments for unmanned ground vehicle development and testing, a system includes stored program instructions, the stored program instructions configured to: receive a first simulation scene and a second simulation scene; calculate a first costmap for the first simulation scene and a second costmap for the second simulation scene; calculate a traversability ratio metric comparing the first costmap and the second costmap; calculate a blob size metric comparing the first costmap and the second costmap; calculate an L2 distance between a first plurality of histograms from the first costmap and a second plurality of histograms from the second costmap; and determine whether a first environment of the first simulation scene and a second environment of the second simulation scene are similar based on the traversability ratio metric, the blob size metric, and the L2 distance.
Core Innovation
The invention provides a system and computer-implemented method for comparing simulated environments for unmanned ground vehicle (UGV) development and testing. It utilizes costmaps, which are top-down images of a geographic area comprised of grid cells, with each cell assigned a value indicating the traversability or cost for the UGV. The approach involves generating costmaps from simulation scenes and binning the costs into categories to enable comparison between different simulation environments.
To address the need for quantitative comparison of virtual environments relevant to UGV performance, the system calculates three metrics from the costmaps: a traversability ratio metric, a blob size metric, and an L2 distance between histograms of the binned costs. The traversability ratio metric compares the ratio of costs between bins, the blob size metric uses connected component labeling to measure the area of connected regions within cost bins, and the L2 distance computes the Euclidean distance between the costmap histograms. These metrics provide a quantitative basis for determining the similarity between different simulated environments.
The problem addressed by the invention is the lack of tools for quantitatively comparing virtual environments used for UGV simulation, which is important given the diverse and challenging conditions faced in military and off-road settings. By enabling numerical assessment of environment similarity based on key costmap characteristics that affect UGV behavior, the system ensures testing environments are representative of expected real-world conditions.
Claims Coverage
The patent contains several independent claims that collectively define four main inventive features.
System for comparing simulated environments using costmap-based metrics
A system comprising one or more computer processors, computer readable storage media, and program instructions configured to: - Receive a first simulation scene and a second simulation scene. - Calculate costmaps for both simulation scenes. - Calculate a traversability ratio metric by comparing the costmaps. - Calculate a blob size metric by comparing the costmaps. - Calculate an L2 distance between histograms constructed from the costmaps' binned values. - Determine whether the environments represented by the simulation scenes are within 20% to 40% similarity according to these three metrics. - Use the traversability ratio metric in determining the UGV's navigational path.
Computer-implemented method for comparing simulated environments using costmap-based metrics
A method executed by one or more computer processors involving the following steps: 1. Receiving a first and a second simulation scene. 2. Calculating costmaps for both simulation scenes. 3. Calculating a traversability ratio metric from the two costmaps. 4. Calculating a blob size metric from the two costmaps. 5. Calculating an L2 distance between pluralities of histograms from the two costmaps. 6. Determining whether the simulation scenes are within 20% to 40% similarity using the three metrics, where the traversability ratio metric informs UGV navigation.
System for comparing simulated environments with defined similarity thresholds
A system comprising computer processors, storage media, and instructions to: - Receive two simulation scenes and calculate respective costmaps. - Calculate a traversability ratio metric, a blob size metric, and an L2 distance between histograms. - Determine that two environments are similar if: - The traversability ratio metric between corresponding bins is between 0.5 and 1.5. - The blob size metric difference between average sizes is less than or equal to 25%. - The L2 distance between histograms is less than or equal to 30%. - The traversability ratio metric is used by the UGV to determine a navigational path.
System for comparing high-resolution simulated environments
A system as above in which each simulation scene consists of thousands to millions of pixels, providing detailed resolution in the costmap-based environment comparison process.
In summary, the inventive features include systems and methods for comparing simulated environments for unmanned ground vehicles using a set of quantitative costmap-based metrics, with explicit thresholds for environment similarity and applicability to high-resolution simulated scenes.
Stated Advantages
Enables quantitative assessment of environment similarity based on features relevant to unmanned ground vehicle performance.
Helps ensure that the set of simulation scenes used for testing is representative of a given target environment while maintaining sufficient variety to robustly test UGV systems.
Documented Applications
Comparing simulated environments for unmanned ground vehicle development and testing.
Interested in licensing this patent?