Preference-based MPC calibration

TL;DR

This paper introduces a semi-automated method for calibrating Model Predictive Controllers using human preferences and active learning, reducing the need for explicit performance metrics and demonstrating effective results in case studies.

Contribution

It proposes a novel preference-based calibration approach for MPCs that leverages human input and active learning to find optimal parameters without explicit performance indices.

Findings

Achieves near-optimal MPC calibration with limited experiments

Effectively learns surrogate performance models from human preferences

Demonstrates success in two case studies

Abstract

Automating the calibration of the parameters of a control policy by means of global optimization requires quantifying a closed-loop performance function. As this can be impractical in many situations, in this paper we suggest a semi-automated calibration approach that requires instead a human calibrator to express a preference on whether a certain control policy is "better" than another one, therefore eliminating the need of an explicit performance index. In particular, we focus our attention on semi-automated calibration of Model Predictive Controllers (MPCs), for which we attempt computing the set of best calibration parameters by employing the recently-developed active preference-based optimization algorithm GLISp. Based on the preferences expressed by the human operator, GLISp learns a surrogate of the underlying closed-loop performance index that the calibrator (unconsciously) uses and proposes, iteratively, a new set of calibration parameters to him or her for testing and for comparison against previous experimental results. The resulting semi-automated calibration procedure is tested on two case studies, showing the capabilities of the approach in achieving near-optimal performance within a limited number of experiments.