Fundamental Challenges in Deep Learning for Stiff Contact Dynamics

被引:17
|
作者
Parmar, Mihir [1 ]
Halm, Mathew [1 ]
Posa, Michael [1 ]
机构
[1] Univ Penn, Grasp Lab, Philadelphia, PA 19104 USA
基金
美国国家科学基金会;
关键词
D O I
10.1109/IROS51168.2021.9636383
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Frictional contact has been extensively studied as the core underlying behavior of legged locomotion and manipulation, and its nearly-discontinuous nature makes planning and control difficult even when an accurate model of the robot is available. Here, we present empirical evidence that learning an accurate model in the first place can be confounded by contact, as modern deep learning approaches are not designed to capture this non-smoothness. We isolate the effects of contact's non-smoothness by varying the mechanical stiffness of a compliant contact simulator. Even for a simple system, we find that stiffness alone dramatically degrades training processes, generalization, and data-efficiency. Our results raise serious questions about simulated testing environments which do not accurately reflect the stiffness of rigid robotic hardware. Significant additional investigation will be necessary to fully understand and mitigate these effects, and we suggest several avenues for future study.
引用
收藏
页码:5181 / 5188
页数:8
相关论文
共 50 条
  • [41] Interpretable machine learning: Fundamental principles and 10 grand challenges
    Rudin, Cynthia
    Chen, Chaofan
    Chen, Zhi
    Huang, Haiyang
    Semenova, Lesia
    Zhong, Chudi
    STATISTICS SURVEYS, 2022, 16 : 1 - 85
  • [42] Deep Fundamental Diagram Network for Fast Pedestrian Dynamics Estimation
    Yi, Ruolong
    Ma, Qing
    Song, Weiguo
    Zhang, Jun
    FIRE TECHNOLOGY, 2024, : 3853 - 3881
  • [43] Geometric deep learning: progress, applications and challenges
    Wenming CAO
    Canta ZHENG
    Zhiyue YAN
    Weixin XIE
    Science China(Information Sciences), 2022, 65 (02) : 238 - 240
  • [44] DEEP LEARNING Tackling the challenges of bioimage analysis
    Pelt, Daniel M.
    ELIFE, 2020, 9 : 1 - 3
  • [45] Big Data Deep Learning: Challenges and Perspectives
    Chen, Xue-Wen
    Lin, Xiaotong
    IEEE ACCESS, 2014, 2 : 514 - 525
  • [46] Deep Reinforcement Learning for Economics: Progress and Challenges
    Green, Etan A.
    Plunkett, E. Barry
    ACM SIGECOM EXCHANGES, 2023, 21 (01) : 49 - 53
  • [47] Deep Learning in LncRNAome: Contribution, Challenges, and Perspectives
    Alam, Tanvir
    Al-Absi, Hamada R. H.
    Schmeier, Sebastian
    NON-CODING RNA, 2020, 6 (04) : 1 - 23
  • [48] Graph Deep Learning: State of the Art and Challenges
    Georgousis, Stavros
    Kenning, Michael P.
    Xie, Xianghua
    IEEE ACCESS, 2021, 9 : 22106 - 22140
  • [49] Blockchain for deep learning: review and open challenges
    Shafay, Muhammad
    Ahmad, Raja Wasim
    Salah, Khaled
    Yaqoob, Ibrar
    Jayaraman, Raja
    Omar, Mohammed
    CLUSTER COMPUTING-THE JOURNAL OF NETWORKS SOFTWARE TOOLS AND APPLICATIONS, 2023, 26 (01): : 197 - 221
  • [50] Graph Mining with Deep Learning: challenges and pitfals
    Appel, Ana Paula
    2019 IEEE INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM WORKSHOPS (IPDPSW), 2019, : 275 - 275