Controller Synthesis for Autonomous Systems With Deep-Learning Perception Components

被引:0
|
作者
Calinescu, Radu [1 ]
Imrie, Calum [1 ]
Mangal, Ravi [2 ]
Rodrigues, Genaina Nunes [3 ]
Pasareanu, Corina [2 ]
Santana, Misael Alpizar [1 ]
Vazquez, Gricel [1 ]
机构
[1] Univ York, Dept Comp Sci, York YO10 5GH, England
[2] Carnegie Mellon Univ, Moffett Field, CA 94035 USA
[3] Univ Brasilia, Dept Comp Sci, BR-70910900 Brasilia, Brazil
基金
英国工程与自然科学研究理事会;
关键词
Discrete-event controller synthesis; Markov model; deep neural network; uncertainty quantification; neuro-symbolic AI; VERIFICATION; MODELS;
D O I
10.1109/TSE.2024.3385378
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
We present DeepDECS, a new method for the synthesis of correct-by-construction software controllers for autonomous systems that use deep neural network (DNN) classifiers for the perception step of their decision-making processes. Despite major advances in deep learning in recent years, providing safety guarantees for these systems remains very challenging. Our controller synthesis method addresses this challenge by integrating DNN verification with the synthesis of verified Markov models. The synthesised models correspond to discrete-event software controllers guaranteed to satisfy the safety, dependability and performance requirements of the autonomous system, and to be Pareto optimal with respect to a set of optimisation objectives. We evaluate the method in simulation by using it to synthesise controllers for mobile-robot collision limitation, and for maintaining driver attentiveness in shared-control autonomous driving.
引用
收藏
页码:1374 / 1395
页数:22
相关论文
共 50 条
  • [1] Perception Computing-Aware Controller Synthesis for Autonomous Systems
    Hobbs, Clara
    Roy, Debayan
    Duggirala, Parasara Sridhar
    Smith, F. Donelson
    Samii, Soheil
    Anderson, James H.
    Chakraborty, Samarjit
    PROCEEDINGS OF THE 2021 DESIGN, AUTOMATION & TEST IN EUROPE CONFERENCE & EXHIBITION (DATE 2021), 2021, : 457 - 462
  • [2] Deep-Learning Tracking for Autonomous Flying Systems Under Adversarial Inputs
    Garcia Carrillo, Luis Rodolfo
    Vamvoudakis, Kyriakos G.
    IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS, 2020, 56 (02) : 1444 - 1459
  • [3] Hardware acceleration technology for deep-learning in autonomous vehicles
    Oh, Chang-song
    Yoon, Jong-min
    2019 IEEE INTERNATIONAL CONFERENCE ON BIG DATA AND SMART COMPUTING (BIGCOMP), 2019, : 686 - 688
  • [4] Improving Testing of Deep-Learning Systems
    Deokuliar, Harsh
    Sangwan, Raghvinder S.
    Badr, Yoaukim
    Srinivasan, Satish M.
    COMMUNICATIONS OF THE ACM, 2024, 67 (03) : 44 - 48
  • [5] Towards Dependable Autonomous Systems Based on Bayesian Deep Learning Components
    Arnez, Fabio
    Espinoza, Huascar
    Radermacher, Ansgar
    Terrier, Francois
    2022 18TH EUROPEAN DEPENDABLE COMPUTING CONFERENCE (EDCC 2022), 2022, : 65 - 72
  • [6] Improving Testing of Deep-learning Systems
    Deokuliar H.
    Sangwan R.S.
    Badr Y.
    Srinivasan S.M.
    Queue, 2023, 21 (05):
  • [7] Deep learning-based perception systems for autonomous driving: A comprehensive survey
    Wen, Li-Hua
    Jo, Kang-Hyun
    NEUROCOMPUTING, 2022, 489 : 255 - 270
  • [8] Deep-learning based autonomous-exploration for UAV navigation
    Zhao, Yumin
    Zhang, Jianlei
    Zhang, Chunyan
    KNOWLEDGE-BASED SYSTEMS, 2024, 297
  • [9] A deep-learning framework for human perception of abstract art composition
    Lelievre, Pierre
    Neri, Peter
    JOURNAL OF VISION, 2021, 21 (05):
  • [10] LaneScanNET: A deep-learning approach for simultaneous detection of obstacle-lane states for autonomous driving systems
    Perumal, P. Shunmuga
    Wang, Yong
    Sujasree, M.
    Tulshain, Shobhit
    Bhutani, Saksham
    Suriyah, M. Kiruthik
    Raju, V. Umesh Kumar
    EXPERT SYSTEMS WITH APPLICATIONS, 2023, 233