Priority-Driven Constraints Softening in Safe MPC for Perturbed Systems
Artikel i vetenskaplig tidskrift, 2025
This letter presents a safe model predictive control framework designed to guarantee the satisfaction of hard safety constraints, for perturbed dynamical systems. Safety is guaranteed by softening the constraints selected on a priority basis from a subset of constraints defined by the designer. Since such an online selection is the result of an auxiliary optimization problem, its computational overhead is alleviated by off-line learning its approximated solution, rather than solving it exactly online. Simulation results, obtained from an automated driving application, show that the proposed approach provides guarantees of collision-avoidance hard constraints despite the unpredicted behaviors of the surrounding environment.
Softening
Model predictive control
Simulation
Control systems
Predictive control
Autonomous vehicles
Dynamical systems
constrained control
Optimization
Computational complexity
uncertain systems
Uncertain systems
Safety
safety critical
Författare
Yingshuai Quan
Chalmers, Elektroteknik, System- och reglerteknik
Mohammad Jeddi
Universita Degli Studi Di Modena E Reggio Emilia
Politecnico di Bari
Francesco Prignoli
Universita Degli Studi Di Modena E Reggio Emilia
Paolo Falcone
Chalmers, Elektroteknik, System- och reglerteknik
IEEE Control Systems Letters
24751456 (eISSN)
Vol. 9 1069-10745G för Uppkopplade Autonoma Fordon i Komplexa Stadsmiljöer
VINNOVA (2018-05005), 2019-04-01 -- 2023-03-31.
Ämneskategorier (SSIF 2025)
Datavetenskap (datalogi)
Datorsystem
Reglerteknik
DOI
10.1109/LCSYS.2025.3580494