RAT iLQR: A Risk Auto-Tuning Controller to Optimally Account for Stochastic Model Mismatch

TL;DR

This paper introduces RAT iLQR, a control algorithm that robustly handles distributional mismatches in stochastic environments by dynamically adjusting risk sensitivity, demonstrated in collision avoidance with uncertain human motion.

Contribution

It proposes a novel nonlinear MPC framework for distributionally robust control with online risk-sensitivity adjustment, addressing imperfect distribution knowledge in stochastic environments.

Findings

Effective in dynamic collision avoidance with uncertain human motion

Automatically adjusts risk sensitivity in real-time

Demonstrates robustness against distributional mismatches

Abstract

Successful robotic operation in stochastic environments relies on accurate characterization of the underlying probability distributions, yet this is often imperfect due to limited knowledge. This work presents a control algorithm that is capable of handling such distributional mismatches. Specifically, we propose a novel nonlinear MPC for distributionally robust control, which plans locally optimal feedback policies against a worst-case distribution within a given KL divergence bound from a Gaussian distribution. Leveraging mathematical equivalence between distributionally robust control and risk-sensitive optimal control, our framework also provides an algorithm to dynamically adjust the risk-sensitivity level online for risk-sensitive control. The benefits of the distributional robustness as well as the automatic risk-sensitivity adjustment are demonstrated in a dynamic collision avoidance scenario where the predictive distribution of human motion is erroneous.