The Southwest Research Institute (SwRI) has developed a 3D simulation tool which can test automated vehicles in virtual off-road environments modeled after real-world scenarios. The announcement comes as expansion of SwRI’s investment into software-in-the-loop solutions to test connected and automated vehicles (CAVs) in traffic-filled or off-road scenarios.
The simulated environment – or 3D software loop – enables an infinite amount of scenarios to be tested to evaluate a vehicle, something that wouldn’t be cost effective in the real world but can instead be conducted in the virtual world. SwRI’s tech is stated to meet the US Department of Defense’s requirements for modeling and simulation tools and will be used to accelerate the development of autonomous, unmanned ground vehicles (UGVs).
Using internal funding, SwRI developed the technology with custom algorithms, off-the-shelf software, open-source tools and public map data. The result was a Simulation Scene Adjustment Tool with a 3D video game-style interface to test virtual ground vehicles on off-road terrain. The simulator can also be used to create digital twins.
“Simulation with the digital twin is crucial for UGV testing and development,” said Joe Auchter, an engineer who led the research for SwRI’s Intelligent Systems Division. “Our Simulation Scene Adjustment Tool allows a user to push UGVs and AVs to the limit and explore ‘what if?’ scenarios in a variety of simulated environments more rapidly, safely and cost effectively than if all this testing was done in the real world.”
The institute’s simulator includes a graphics engine, dynamics engine, vehicle modeling tools, vehicle terrain interaction models and plug-ins to communicate with an autonomy software stack. It builds scenes with elevation maps captured from geographic information system (GIS) data and graphically renders topographical features in 3D.
The first round of research incorporated digital elevation models (DEMs) from aerial scans conducted by the San Antonio River Authority and other government agencies.
“We developed algorithms to perturb DEM and GIS data in user-configurable ways that generate synthetic environments,” Auchter said. “This allows for testing of new algorithms and techniques in simulation, building numerous test environments that share certain relevant characteristics with a real geo-specific location where vehicles will eventually operate.”
SwRI’s machine learning algorithms can also simulate computer vision and sensing outputs for lidar, radar, cameras and GPS among other systems to perceive scene objects, movements and position when calculating driving responses. A dynamics engine simulates forces caused by gravity and motion as a vehicle model moves through an environment. Furthermore, the simulated vehicle can have its weight, speed, horsepower, center of gravity and others attributes programmed.
“If you look at field testing of automated vehicles, there are simply not enough miles or novel situations that you can throw at a vehicle to encounter all the edge cases for sensors and software,” said Jerry Towler, assistant director of SwRI’s Robotics Department. “Modeling and simulation help test AVs and advanced driver assistance systems (ADAS) to enhance safety and ensure capability before and alongside deployment into real-world testing environments.”
At present the simulator is being used to support military and civilian client applications.
