PhD project Physics-based simulation is essential for developing and evaluating machines that operate in complex physical environments. Simulation Intelligence explores how artificial intelligence can automate the creation, execution, and analysis of simulations, making it possible to investigate more scenarios with less manual effort and better use of computational resources.
The project develops AI-based methods for automating and improving physics-based simulation workflows. The research focuses on generating relevant simulation scenarios, select and develop appropriate models models and computational effort, and analysing simulation outcomes. The main applications concern off-road and earthmoving machinery interacting with complex and deformable terrain.
Physics-based simulation enables engineers and researchers to investigate how machines behave and interact with their surroundings without relying exclusively on physical experiments. This is particularly valuable for off-road and earthmoving machinery, where interactions between machines, terrain, and materials are complex and operating conditions can vary considerably.
Despite advances in simulation technology, creating and using realistic simulation environments remains a demanding process. Simulation scenarios must be constructed, models and parameters selected, computational resources allocated, and results evaluated. Exploring many different operating conditions can require substantial manual effort and computing time.
Simulation Intelligence investigates how artificial intelligence can automate and improve these processes. The research addresses methods for generating and selecting relevant simulation scenarios, adapting the level of simulation detail and computational effort, and automatically analysing simulation outcomes. By learning from simulation results, intelligent methods may also help determine which simulations are most valuable to perform next.
The overall aim is to make physics-based simulation more systematic, efficient, and informative. The research primarily targets mobile machines operating in complex terrain, including forestry, construction, and earthmoving equipment, but the methods may also be applicable to other robotic and mechanical systems.
For further information, visit our homepage!
