Addressing Behavior Model Inaccuracies for Safe Motion Control in Uncertain Dynamic Environments
Sung, Minjun
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/129560
Description
Title
Addressing Behavior Model Inaccuracies for Safe Motion Control in Uncertain Dynamic Environments
Author(s)
Sung, Minjun
Issue Date
2025-04-24
Director of Research (if dissertation) or Advisor (if thesis)
Hovakimyan, Naira
Department of Study
Mechanical Sci & Engineering
Discipline
Mechanical Engineering
Degree Granting Institution
University of Illinois Urbana-Champaign
Degree Name
M.S.
Degree Level
Thesis
Keyword(s)
Motion Control
Robust/Adaptive Control
Planning under Uncertainty
Collision Avoidance
Abstract
Uncertainties in the environment and inaccuracies in behavior models critically affect the safety and reliability of autonomous systems in dynamic environment. These inaccuracies compromise the estimation of a dynamic obstacle’s state, leading to biased estimates and shifts in the predicted trajectory distributions. Such prediction errors, if unaddressed, may result in violations of safety constraints and degraded control performance. To address these challenges, we propose a novel framework called SIED-MPC (Simultaneous Input-Estimation and Distributionally robust Model Predictive Control), which unifies Simultaneous State and Input Estimation (SSIE) with Distributionally Robust Model Predictive Control (DR-MPC) through an adaptive model confidence evaluation scheme. Unlike conventional estimation techniques that assume access to accurate behavior models or treat prediction as an isolated module, our SSIE formulation jointly estimates both the obstacle’s state and the input gap—the discrepancy between predicted and actual control inputs—thus correcting for behavior model errors in real-time. This input gap serves as a quantitative proxy for model confidence, which is used to dynamically adjust the size of the ambiguity set in the DR-MPC formulation via a Wasserstein-based uncertainty radius. By integrating this feedback-driven adaptivity into the control pipeline, SIED-MPC systematically accounts for both estimation bias and distributional shift, ensuring safe operation with minimal conservatism. The proposed framework is evaluated in realistic autonomous driving simulations using CARLA. Our method demonstrates superior collision avoidance performance, lower constraint violation rates, and improved computational efficiency.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
