Learning Dynamics from Infrequent Output Measurements for Uncertainty-Aware Optimal Control
This addresses the challenge of reliable control in systems with limited sensing, such as medical applications like diabetes management, but is incremental as it builds on existing Bayesian and control methods.
The paper tackled the problem of optimal control for nonlinear systems with unknown dynamics and infrequent, noisy measurements by developing a Bayesian method that updates a prior over dynamics and latent states using a targeted sampler, then solves a scenario-based control problem; it was validated on a Type 1 diabetes model, showing reliable regulation.
Reliable optimal control is challenging when the dynamics of a nonlinear system are unknown and only infrequent, noisy output measurements are available. This work addresses this setting of limited sensing by formulating a Bayesian prior over the continuous-time dynamics and latent state trajectory in state-space form and updating it through a targeted marginal Metropolis-Hastings sampler equipped with a numerical ODE integrator. The resulting posterior samples are used to formulate a scenario-based optimal control problem that accounts for both model and measurement uncertainty and is solved using standard nonlinear programming methods. The approach is validated in a numerical case study on glucose regulation using a Type 1 diabetes model.