Optimal feedback controls for nonlinear systems are characterized by
the solutions to a Hamilton Jacobi Bellmann (HJB) equation. In the
deterministic case, this is a first order hyperbolic equation. Its
dimension is that of the state-space of the nonlinear system. Thus,
solving the HJB equation is a formidable task.
In practice, optimal feedback controls are frequently based on
linearization and subsequent treatment by efficient Riccati solvers.
This can be effective, but it is a local procedure, and it may fail or
lead to erroneous results.
In this talk, I describe three approaches. The first one is based on
Newton steps applied to the HJB equation. Combined with tensor
calculus this allows to approximately solve HJB equations up to
dimension 100. Results are demonstrated for the control of discretized
Fokker Planck equations. The second approach is a data driven
technique, which approximated the HJB equation and its gradient from
an ensemble of open loop solves.The third technique circumvents the
direct solution of the HJB equation. Rather a neural network is
trained by means of a succinctly chosen ansatz.
This work relies on collaborations with B. Azmi, S. Dolgov,
D. Kalise, L. Pfeiffer, and D. Walter.
|