An essential activity carried out within the ARNL is basic research. On the one hand, basic research is indispensable in solving various complex application tasks, and on the other hand, it ensures an international reputation through the internationally published leading and trend-aligned novel research results. The main topics of basic research within the ARNL are modelling, model-reduction and model-identification, and the conrol of adaptive, robust, as well as distributed and/or networked systems.
Among the control methods, the linear variable parameter (LPV) control methods, the model predictive control (MPC) methods, and the reinforcement learning methods should be mentioned; but methods facilitating real-time implementations, (e.g., in conjunction with trajectory planning, and environmental sensor fusion) are also investigated by the ARNL basic research team.
It is a challenge to select suitable performance specifications that meet robustness requirements. In case of probabilistic relaxation, which is used to increase control efficiency, compliance with only the vast majority of cases is required. The case-based approach makes it possible to handle situations where the standard approaches cannot be applied (e.g., due to the exponential increase in their computation times). However, examining the robustness of such methods is a significant research challenge.
Basic research in modelling, model-reduction and model-identification. LPV and MPC methods for trajectory planning, as well as for longitudinal and lateral vehicle motion control.
Optimal control design via machine learning. Supporting complex, computationally demanding nonlinear optimization tasks with machine learning approaches to facilitate straightforward real-time implementations.
Measurement and data processing of vehicle dynamics and traffic features during data collection campaigns. Locating and acquiring the available public data sources. Investigating the processing possibilities in conjunction with the gathered large volume data that is stored and managed in a cloud-based database.