A sensitivity based methodology is presented for choosing the best possible gains parametrization in a SDRE (State Riccati Dependent Equation) controller. Both direct and adjoint sensitivity methods are used, together with a dynamic inversion of the system, in order to optimize the performances of that kind of controller. Special attention is given to the choice of state and control matrices defining the criteria to be minimized by the Riccati equation. It is described how various deterministic and statistical sensitivity approaches may help to parametrize in a suboptimal way the matrices. A feed-forward/feedback steering control strategy for a vehicle has been chosen as application. Obstacle avoiding simulation results are shown and compared with a similar strategy where the feedback block is obtained by adaptive gains. The performance improvements are shown in terms of path tracking when compared with the on-line optimization.
Sensitivity based gains choice for a SDRE vehicle steering control strategy
Empfindlichkeitsbasierte Verstärkungseinstellung für die Fahrzeuglenkung mit der Lösung einer zustandsabhängigen Riccati-Gleichung
2006
7 Seiten, 13 Tabellen
Aufsatz (Konferenz)
Englisch
SDRE control strategy applied to a nonlinear robotic including drive motor
American Institute of Physics | 2014
|SDRE Trajectory Tracking Control for a Hovercraft Autonomous Vehicle
Springer Verlag | 2018
|Stability-based SDRE controller for spacecraft momentum management
Online Contents | 2013
|