Issue 39 Unmanned Systems Technology August/September 2021 Maritime Robotics Mariner l Simulation tools focus l MRS MR-10 and MR-20 l UAVs insight l HFE International GenPod l Exotec Skypod l Autopilots focus l Aquaai Mazu

42 Focus | Simulation tools focus on specific actors in the simulation rather than the entire scene, and can boost the path-planning algorithms in an autonomous vehicle. To evaluate the technology, a virtual model based on real data collected during test drives with a driverless car was created. The simulations showed that the safety module does not lead to any loss of performance in driving behaviour, and that the forecast calculations are correct, preventing accidents and halting the vehicle safely in an emergency. Conclusion Processing power has allowed simulation engines to handle more vehicles with higher fidelity, but it is the evolution of the industry standards for performing the simulations and the standards for the simulation engines themselves that are proving to be key. Simulation has come a long way since the first use of the Unity and Unreal gaming engines to create virtual worlds to test a vehicle. The limitations of the gaming engines, coupled with the development of high-fidelity digital twin models of the sensors, has required a new generation of simulation engines that can handle much higher levels of complexity. These can now run models of multiple independent vehicles in a virtual environment with accurate physics representation of challenging conditions including rain, snow or ice. This allows the performance of individual vehicles and the complex interactions of multiple sensors to be evaluated and analysed for the edge and corner cases, which are hard to test in the real world. Simulation confers several other advantages though. The digital twin models are not just for the development cycle. Running a digital twin through the lifetime of a vehicle with predictive maintenance machine learning algorithms can provide early identification of problems, allowing repairs and replacements to be performed on a managed basis. This boosts the safety of the vehicle in whichever realm it is operating, and just as new statistical safety standards such as SOTIF have emerged for the development cycle, so they will emerge for safety throughout the life cycle of the design. These simulation algorithms are also making their way into the vehicle itself, to enhance safety during operation. Running simplified models on the vehicle can provide predictions of the likely movement of surrounding objects and link to the path- planning algorithms to find a safe choice. This will further increase the processing requirements in the vehicle. Acknowledgements The author would like to thank Gilles Gallee at Ansys, Szabolcs Janky and Gergely Juhasz at AImotive, Becky Ham at the Department of Aeronautics and Astronautics at MIT, and Prof Matthias Althoff at the Munich School of Robotics and Machine Intelligence at the Technical University of Munich for their help with researching this article. August/September 2021 | Unmanned Systems Technology FRANCE AVsimulation +33 1 4694 9780 www.avsimulation.fr GERMANY Siemens (Mentor) – www.plm.automation.siemens.com HUNGARY AImotive +36 1 7707 201 www.aimotive.com INDIA Wipro +91 80 4682 7999 www.wipro.com ISRAEL Cognata +1 855 500 0217 www.cognata.com USA Ansys +1 724 746 3304 www.ansys.com NI (Monodrive) – www.monodrive.io Nvidia +1 408 486 2500 www.nvidia.com Opal-RT Technologies +1 514 935 2323 www.opal-rt.com TotalSim +1 614 255 7426 www.totalsim.us Voyage (Cruise) +1 206 623 1986 www.voyage.auto NO LOCATION (OPEN SOURCE) CARLA – www.carla.org Some examples of simulation tools suppliers