Uncertainty-Aware Data Aggregation for Deep Imitation Learning

被引:0
|
作者
Cui, Yuchen [1 ]
Isele, David [2 ]
Niekum, Scott [1 ]
Fujimura, Kikuo [2 ]
机构
[1] Univ Texas Austin, Austin, TX 78712 USA
[2] Honda Res Inst USA, 375 Ravendale Dr, Mountain View, CA 94043 USA
关键词
NEURAL-NETWORKS;
D O I
10.1109/icra.2019.8794025
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Estimating statistical uncertainties allows autonomous agents to communicate their confidence during task execution and is important for applications in safety-critical domains such as autonomous driving. In this work, we present the uncertainty-aware imitation learning (UAIL) algorithm for improving end-to-end control systems via data aggregation. UAIL applies Monte Carlo Dropout to estimate uncertainty in the control output of end-to-end systems, using states where it is uncertain to selectively acquire new training data. In contrast to prior data aggregation algorithms that force human experts to visit sub-optimal states at random, UAIL can anticipate its own mistakes and switch control to the expert in order to prevent visiting a series of sub-optimal states. Our experimental results from simulated driving tasks demonstrate that our proposed uncertainty estimation method can be leveraged to reliably predict infractions. Our analysis shows that UAIL outperforms existing data aggregation algorithms on a series of benchmark tasks.
引用
收藏
页码:761 / 767
页数:7
相关论文
共 50 条
  • [1] Uncertainty-Aware Imitation Learning using Kernelized Movement Primitives
    Silverio, Joao
    Huang, Yanlong
    Abu-Dakka, Fares J.
    Rozo, Leonel
    Caldwell, Darwin G.
    [J]. 2019 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2019, : 90 - 97
  • [2] Uncertainty-Aware Policy Sampling and Mixing for Safe Interactive Imitation Learning
    Diaz, Manfred
    Fevens, Thomas
    Paull, Liam
    [J]. 2021 18TH CONFERENCE ON ROBOTS AND VISION (CRV 2021), 2021, : 72 - 78
  • [3] Uncertainty-Aware Deep Learning Based Deformable Registration
    Grigorescu, Irina
    Uus, Alena
    Christiaens, Daan
    Cordero-Grande, Lucilio
    Hutter, Jana
    Batalle, Dafnis
    Edwards, A. David
    Hajnal, Joseph V.
    Modat, Marc
    Deprez, Maria
    [J]. UNCERTAINTY FOR SAFE UTILIZATION OF MACHINE LEARNING IN MEDICAL IMAGING, AND PERINATAL IMAGING, PLACENTAL AND PRETERM IMAGE ANALYSIS, 2021, 12959 : 54 - 63
  • [4] Uncertainty-aware deep learning for monitoring and fault diagnosis from synthetic data
    Das, Laya
    Gjorgiev, Blazhe
    Sansavini, Giovanni
    [J]. RELIABILITY ENGINEERING & SYSTEM SAFETY, 2024, 251
  • [5] Training Uncertainty-Aware Classifiers with Conformalized Deep Learning
    Einbinder, Bat-Sheva
    Romano, Yaniv
    Sesia, Matteo
    Zhou, Yanfei
    [J]. ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 35, NEURIPS 2022, 2022,
  • [6] Uncertainty-aware autonomous sensing with deep reinforcement learning
    Murad, Abdulmajid
    Kraemer, Frank Alexander
    Bach, Kerstin
    Taylor, Gavin
    [J]. FUTURE GENERATION COMPUTER SYSTEMS-THE INTERNATIONAL JOURNAL OF ESCIENCE, 2024, 156 : 242 - 253
  • [7] Uncertainty-Aware Data Augmentation for Offline Reinforcement Learning
    Su, Yunjie
    Kong, Yilun
    Wang, Xueqian
    [J]. 2023 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS, IJCNN, 2023,
  • [8] Uncertainty-Aware Multiview Deep Learning for Internet of Things Applications
    Xu, Cai
    Zhao, Wei
    Zhao, Jinglong
    Guan, Ziyu
    Song, Xiangyu
    Li, Jianxin
    [J]. IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, 2023, 19 (02) : 1456 - 1466
  • [9] Uncertainty-Aware Action Advising for Deep Reinforcement Learning Agents
    Da Silva, Felipe Lena
    Hernandez-Leal, Pablo
    Kartal, Bilal
    Taylor, Matthew E.
    [J]. THIRTY-FOURTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THE THIRTY-SECOND INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE AND THE TENTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2020, 34 : 5792 - 5799
  • [10] Exploration in Online Advertising Systems with Deep Uncertainty-Aware Learning
    Du, Chao
    Gao, Zhifeng
    Yuan, Shuo
    Gao, Lining
    Li, Ziyan
    Zeng, Yifan
    Zhu, Xiaoqiang
    Xu, Jian
    Gai, Kun
    Lee, Kuang-Chih
    [J]. KDD '21: PROCEEDINGS OF THE 27TH ACM SIGKDD CONFERENCE ON KNOWLEDGE DISCOVERY & DATA MINING, 2021, : 2792 - 2801