Discovering Generalizable Skills via Automated Generation of Diverse Tasks

被引:0
|
作者
Fang, Kuan [1 ]
Zhu, Yuke [2 ,3 ]
Savarese, Silvio [1 ]
Li Fei-Fei [1 ]
机构
[1] Stanford Univ, Stanford, CA 94305 USA
[2] UT Austin, Austin, TX USA
[3] Nvidia, Santa Clara, CA USA
关键词
REINFORCEMENT; OBJECTS;
D O I
暂无
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
The learning efficiency and generalization ability of an intelligent agent can be greatly improved by utilizing a useful set of skills. However, the design of robot skills can often be intractable in real-world applications due to the prohibitive amount of effort and expertise that it requires. In this work, we introduce Skill Learning In Diversified Environments (SLIDE), a method to discover generalizable skills via automated generation of a diverse set of tasks. As opposed to prior work on unsupervised discovery of skills which incentivizes the skills to produce different outcomes in the same environment, our method pairs each skill with a unique task produced by a trainable task generator. To encourage generalizable skills to emerge, our method trains each skill to specialize in the paired task and maximizes the diversity of the generated tasks. A task discriminator defined on the robot behaviors in the generated tasks is jointly trained to estimate the evidence lower bound of the diversity objective. The learned skills can then be composed in a hierarchical reinforcement learning algorithm to solve unseen target tasks. We demonstrate that the proposed method can effectively learn a variety of robot skills in two tabletop manipulation domains. Our results suggest that the learned skills can effectively improve the robot's performance in various unseen target tasks compared to existing reinforcement learning and skill learning methods.
引用
收藏
页数:12
相关论文
共 50 条
  • [21] MedRAT: Unpaired Medical Report Generation via Auxiliary Tasks
    Hirsch, Elad
    Dawidowicz, Gefen
    Tal, Ayellet
    COMPUTER VISION - ECCV 2024, PT LXXI, 2025, 15129 : 18 - 35
  • [22] Unifying Vision-and-Language Tasks via Text Generation
    Cho, Jaemin
    Lei, Jie
    Tan, Hao
    Bansal, Mohit
    INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 139, 2021, 139
  • [23] Transferring Skills to Robots for Tasks with Cyclic Motions via Dynamical Systems Approach
    Vakanski, Aleksandar
    Janabi-Sharifi, Farrokh
    Mantegh, Iraj
    2012 INTERNATIONAL SYMPOSIUM ON OPTOMECHATRONIC TECHNOLOGIES (ISOT), 2012,
  • [24] Discovering diverse soliton solutions in the modified Schrödinger's equation via innovative approaches
    Ke, Shanwen
    Shateyi, S.
    Alqahtani, Salman A.
    Alqahtani, Nouf F.
    RESULTS IN PHYSICS, 2024, 57
  • [25] Supporting unit test generation via automated isolation
    Honfi D.
    Micskei Z.
    Periodica polytechnica Electrical engineering and computer science, 2017, 61 (02): : 116 - 131
  • [26] Automated Hypotheses Generation via Combinatorial Causal Optimization
    Pietrantuono, Roberto
    2021 IEEE CONGRESS ON EVOLUTIONARY COMPUTATION (CEC 2021), 2021, : 399 - 407
  • [27] Automated Test Oracle Generation via Denotational Semantics
    Guo, Hai-Feng
    Cao, Liang
    Song, Yushu
    Qiu, Zongyan
    2014 14TH INTERNATIONAL CONFERENCE ON QUALITY SOFTWARE (QSIC 2014), 2014, : 139 - 144
  • [28] Automated Library Generation and Serendipity Quantification Enables Diverse Discovery in Coordination Chemistry
    Kowalski, Daniel J.
    MacGregor, Catriona M.
    Long, De-Liang
    Bell, Nicola L.
    Cronin, Leroy
    JOURNAL OF THE AMERICAN CHEMICAL SOCIETY, 2023, 145 (04) : 2332 - 2341
  • [29] Automatic Generation of Flexible Plans via Diverse Temporal Planning
    Amitai, Yotam
    Taitler, Ayal
    Karpas, Erez
    THIRTY-FIFTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THIRTY-THIRD CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE AND THE ELEVENTH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2021, 35 : 6049 - 6057
  • [30] Emu Edit: Precise Image Editing via Recognition and Generation Tasks
    Sheynin, Shelly
    Polyak, Adam
    Singer, Uriel
    Kirstain, Yuval
    Zohar, Amit
    Ashual, Oron
    Parikh, Devi
    Taigman, Yaniv
    2024 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2024, 2024, : 8871 - 8879