Online Data-Driven Safety Certification for Systems Subject to Unknown Disturbances

被引:0
|
作者
Rober, Nicholas [1 ]
Mahesh, Karan [2 ]
Paine, Tyler M. [3 ,4 ]
Greene, Max L. [2 ]
Lee, Steven [2 ]
Monteiro, Sildomar T. [2 ]
Benjamin, Michael R. [3 ]
How, Jonathan P. [1 ]
机构
[1] MIT, Aerosp Controls Lab, Cambridge, MA 02139 USA
[2] Aurora Flight Sci, Cambridge, MA USA
[3] MIT, Marine Auton Lab, Cambridge, MA USA
[4] Woods Hole Oceanog Inst, Woods Hole, MA 02543 USA
关键词
KALMAN FILTER;
D O I
10.1109/ICRA57147.2024.10610163
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Deploying autonomous systems in safety critical settings necessitates methods to verify their safety properties. This is challenging because real-world systems may be subject to disturbances that affect their performance, but are unknown a priori. This work develops a safety-verification strategy wherein data is collected online and incorporated into a reachability analysis approach to check in real-time that the system avoids dangerous regions of the state space. Specifically, we employ an optimization-based moving horizon estimator (MHE) to characterize the disturbance affecting the system, which is incorporated into an online reachability calculation. Reachable sets are calculated using a computational graph analysis tool to predict the possible future states of the system and verify that they satisfy safety constraints. We include theoretical arguments proving our approach generates reachable sets that bound the future states of the system, as well as numerical results demonstrating how it can be used for safety verification. Finally, we present results from hardware experiments demonstrating our approach's ability to perform online reachability calculations for an unmanned surface vehicle subject to currents and actuator failures.
引用
收藏
页码:9939 / 9945
页数:7
相关论文
共 50 条
  • [1] Data-Driven Strategy Synthesis for Stochastic Systems with Unknown Nonlinear Disturbances
    Gracia, Ibon
    Boskos, Dimitris
    Laurenti, Luca
    Lahijanian, Morteza
    6TH ANNUAL LEARNING FOR DYNAMICS & CONTROL CONFERENCE, 2024, 242 : 1633 - 1645
  • [2] Data-Driven Robust Control of Unknown MIMO Nonlinear System Subject to Input Saturations and Disturbances
    Wang, Li
    Gong, Huajun
    Liu, Chunsheng
    MATHEMATICAL PROBLEMS IN ENGINEERING, 2017, 2017
  • [3] Online learning of data-driven controllers for unknown switched linear systems
    Rotulo, Monica
    De Persis, Claudio
    Tesi, Pietro
    AUTOMATICA, 2022, 145
  • [4] Data-Driven Reinforcement Learning Design for Multi-agent Systems with Unknown Disturbances
    Zhong, Xiangnan
    Ni, Zhen
    2018 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), 2018,
  • [5] SAFETY IN THE FACE OF UNKNOWN UNKNOWNS: ALGORITHM FUSION IN DATA-DRIVEN ENGINEERING SYSTEMS
    Kshetry, Nina
    Varshney, Lav R.
    2019 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2019, : 8162 - 8166
  • [6] Data-Driven H∞ Control for Unknown Linear Time-Invariant Systems with Bounded Disturbances
    Hu, Kaijian
    Liu, Tao
    2022 IEEE 61ST CONFERENCE ON DECISION AND CONTROL (CDC), 2022, : 1423 - 1428
  • [7] Online Stochastic Optimization for Unknown Linear Systems: Data-Driven Controller Synthesis and Analysis
    Bianchin, Gianluca
    Vaquero, Miguel
    Cortes, Jorge
    Dall'Anese, Emiliano
    IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 2024, 69 (07) : 4411 - 4426
  • [8] Online Data-Driven Adaptive Control for Unknown Linear Time-Varying Systems
    Liu, Shenyu
    Chen, Kaiwen
    Eising, Jaap
    2023 62ND IEEE CONFERENCE ON DECISION AND CONTROL, CDC, 2023, : 8775 - 8780
  • [9] Online adaptive data-driven control for unknown nonlinear systems with constrained-input
    Xi'an University of Architecture and Technology, Xi'an, China
    Int. Conf. Cyber-Energy Syst. Intell. Energy, ICCSIE, 1600,
  • [10] Data-Driven Synthesis of Safety Controllers for Partially-Observable Systems with Unknown Models
    Jahanshahi, Niloofar
    Zamani, Majid
    2023 62ND IEEE CONFERENCE ON DECISION AND CONTROL, CDC, 2023, : 1052 - 1057