Control-Tree Optimization: an approach to MPC under discrete Partial Observability

TL;DR

This paper introduces a control-tree based MPC approach for environments with discrete partial observability, optimizing policies based on probabilistic belief states to improve performance and robustness.

Contribution

It proposes a novel control-tree optimization method that accounts for belief states, enabling scalable, real-time MPC under discrete partial observability.

Findings

Effective in linear and non-linear MPC with constraints

Parallel optimization enhances scalability

Demonstrates real-time feasibility and benefits over classical MPC

Abstract

This paper presents a new approach to Model Predictive Control for environments where essential, discrete variables are partially observed. Under this assumption, the belief state is a probability distribution over a finite number of states. We optimize a \textit{control-tree} where each branch assumes a given state-hypothesis. The control-tree optimization uses the probabilistic belief state information. This leads to policies more optimized with respect to likely states than unlikely ones, while still guaranteeing robust constraint satisfaction at all times. We apply the method to both linear and non-linear MPC with constraints. The optimization of the \textit{control-tree} is decomposed into optimization subproblems that are solved in parallel leading to good scalability for high number of state-hypotheses. We demonstrate the real-time feasibility of the algorithm on two examples and show the benefits compared to a classical MPC scheme optimizing w.r.t. one single hypothesis.