Control-Oriented Learning on the Fly
This addresses the challenge of maintaining control in safety-critical scenarios like system failures, though it is incremental as it builds on existing control theory with a new algorithmic approach.
The paper tackles the problem of controlling systems with unknown dynamics, such as after a critical failure, by developing a myopic control strategy that optimizes the immediate trajectory direction using only current information. The proposed algorithm uses small control perturbations to learn local dynamics while ensuring near-optimal movement, with hard suboptimality bounds verified in simulations of a damaged aircraft and a Van der Pol oscillator.
This paper focuses on developing a strategy for control of systems whose dynamics are almost entirely unknown. This situation arises naturally in a scenario where a system undergoes a critical failure. In that case, it is imperative to retain the ability to satisfy basic control objectives in order to avert an imminent catastrophe. A prime example of such an objective is the reach-avoid problem, where a system needs to move to a certain state in a constrained state space. To deal with limitations on our knowledge of system dynamics, we develop a theory of myopic control. The primary goal of myopic control is to, at any given time, optimize the current direction of the system trajectory, given solely the information obtained about the system until that time. We propose an algorithm that uses small perturbations in the control effort to learn local dynamics while simultaneously ensuring that the system moves in a direction that appears to be nearly optimal, and provide hard bounds for its suboptimality. We additionally verify the usefulness of the algorithm on a simulation of a damaged aircraft seeking to avoid a crash, as well as on an example of a Van der Pol oscillator.