Reduced Order Modeling of tractor-semitrailer with Response Surface Methodology

Safety is one of the main challenges in vehicle automation, requiring an extensive phase of testing, with vehicle dynamics models that are fundamental for this purpose. The requirement on extensive regression testing and software development requirements raise the need for low-fidelity models with a shorter simulation time with respect to multibody models, the usual models for the simulation of heavy vehicles dynamics.
The thesis, carried out at Volvo Autonomous Solutions, aims to create a low-fidelity model of a tractor-semitrailer dynamics. The model consists of a reduced order model based on a one step ahead prediction of kinematic variables obtained from the multibody model. The prediction is performed through algorithms belonging to the response surface methodology. Finally, the model, obtained as a Functional Mock-up Unit, undergoes a validation phase through the simulation of some manoeuvres and the evaluation of the error between the outputs time trajectories obtained for the low-fidelity model and for the multibody model. The results show the potential of the methodology to create a low-fidelity model that reproduces the dynamics of a truck, in particular with the utilization of response surface models based on a polynomial SVD algorithm.