Collision Avoidance with Stochastic Model Predictive Control for Systems with a Twofold Uncertainty Structure

TL;DR

This paper introduces a novel stochastic model predictive control framework that effectively manages two types of uncertainty—task-level and execution-level—in automated vehicle systems, enhancing robustness and efficiency.

Contribution

The paper presents a combined scenario and analytic stochastic MPC approach specifically designed for twofold uncertainty in automated vehicles, a novel integration not previously addressed.

Findings

Successfully applied in automated vehicle simulations

Improved handling of multi-task and execution uncertainties

Demonstrated robustness and efficiency in uncertain environments

Abstract

Model Predictive Control (MPC) has shown to be a successful method for many applications that require control. Especially in the presence of prediction uncertainty, various types of MPC offer robust or efficient control system behavior. For modeling, uncertainty is most often approximated in such a way that established MPC approaches are applicable for specific uncertainty types. However, for a number of applications, especially automated vehicles, uncertainty in predicting the future behavior of other agents is more suitably modeled by a twofold description: a high-level task uncertainty and a low-level execution uncertainty of individual tasks. In this work, we present an MPC framework that is capable of dealing with this twofold uncertainty. A scenario MPC approach considers the possibility of other agents performing one of multiple tasks, with an arbitrary probability distribution, while an analytic stochastic MPC method handles execution uncertainty within a specific task, based on a Gaussian distribution. Combining both approaches allows to efficiently handle the twofold uncertainty structure of many applications. Application of the proposed MPC method is demonstrated in an automated vehicle simulation study.