Paper presented at ISTVS 2025 | 55th Conference of the International Society for Terrain-Vehicle Systems https://doi.org/10.56884/JCVHRORX
Authors: Gerhard Skoff, Daniel Hassler, Gernot Hasenbichler
Keywords: Autonomous Systems; Validation; Offroad-Mobility; Simulation
Abstract:
To ensure safe and robust autonomous offroad-operation, specific challenges must be managed, including robust terrain-specific obstacle detection. Virtual approaches have the potential to significantly shorten development time and effort. Adequate methods need to consider changing environmental conditions, as e.g. changes of the vegetation during the seasons of the year as well as weather. To manage all those requirements, a modular simulation platform has been developed.
The toolchain consists of six independent and flexible modules, which are all linked together to conduct simulations seamlessly and with best correlation to real world:
Vehicle-specific multibody dynamic simulation module of the vehicle, which can be derived from different multibody dynamic simulation tools, whereby a mathematical soft-soil model applies,
A terrain model, which utilizes a comprehensive set of specifically designed terrain elements, and which undergo AI-supported modifications, e.g. darkness, fog or rain during the validation process,
A sensor module, which can operate with different kinds of active and non-emitting sensors. Sensor selection and arrangement on the vehicle are optimized in the virtual sensor lab,
An autonomous algorithm, which is the system-under-test (SuT) and which gets optimized and validated,
Set of terrain-specific KPI’s, which define application-specific boundaries and qualification criteria,
A terrain-specific scenario database, where ontology-based automated offroad-scenarios are generated.
The sensor model and SuT get integrated into the Vehicle model, and the vehicle is executing the automatically generated, ontology-based testcase-combinations. Application-specific KPI’s are calculated and compared with target values.
Last updated
