SOZIL: Self-Optimal Zero-shot Imitation Learning
文献类型:期刊论文
; | |
作者 | Peng Hao; Tao Lu; Shaowei Cui; Junhang Wei; Shuo Wang; Yinghao Cai |
刊名 | IEEE Trans on Cognitive and Developmental System ; IEEE Trans on Cognitive and Developmental System |
出版日期 | 2021 ; 2021 |
期号 | 1页码:1 |
关键词 | imitation learning imitation learning learning from observation keyframe demonstration learning from observation keyframe demonstration |
英文摘要 | Zero-shot imitation learning has demonstrated its superiority to learn complex robotic tasks with less human participation. Recent studies show convincing performance under the condition that the robot follows the demonstration strictly by the learned inverse model. However, these methods are difficult to achieve satisfactory performance in imitation when the demonstration is suboptimal, and the learning of the learned inverse models is vulnerable to label ambiguity issues. In this paper, we propose Self-Optimal Zero-shot Imitation Learning (SOZIL) to tackle these problems. The contribution of SOZIL is twofold. First, Goal Consistency Loss (GCL) is designed to learn the multi-step goal-conditioned policy from exploration data. By directly using the goal state as supervision, GCL solves the label ambiguity problem caused by trajectory and action diversity. Second, Estimation-based Keyframe Extraction(EKE) is developed to optimize demonstrations. We formulate the keyframe extraction process as a path optimization problem under suboptimal control. By predicting the performance of the learned policy in executing transitions of any two states, EKE creates a directed graph containing all candidate paths and extracts keyframes by solving the graph’s shortest path problem. Furthermore, the proposed method is evaluated with various simulated and real-world robotic manipulating experiments such as cable harness assembly, rope manipulation, and block moving. Experimental results show that SOZIL achieves a superior success rate and manipulation efficiency than baselines Zero-shot imitation learning has demonstrated its superiority to learn complex robotic tasks with less human participation. Recent studies show convincing performance under the condition that the robot follows the demonstration strictly by the learned inverse model. However, these methods are difficult to achieve satisfactory performance in imitation when the demonstration is suboptimal, and the learning of the learned inverse models is vulnerable to label ambiguity issues. In this paper, we propose Self-Optimal Zero-shot Imitation Learning (SOZIL) to tackle these problems. The contribution of SOZIL is twofold. First, Goal Consistency Loss (GCL) is designed to learn the multi-step goal-conditioned policy from exploration data. By directly using the goal state as supervision, GCL solves the label ambiguity problem caused by trajectory and action diversity. Second, Estimation-based Keyframe Extraction(EKE) is developed to optimize demonstrations. We formulate the keyframe extraction process as a path optimization problem under suboptimal control. By predicting the performance of the learned policy in executing transitions of any two states, EKE creates a directed graph containing all candidate paths and extracts keyframes by solving the graph’s shortest path problem. Furthermore, the proposed method is evaluated with various simulated and real-world robotic manipulating experiments such as cable harness assembly, rope manipulation, and block moving. Experimental results show that SOZIL achieves a superior success rate and manipulation efficiency than baselines |
语种 | 英语 ; 英语 |
源URL | [http://ir.ia.ac.cn/handle/173211/47501] |
专题 | 智能机器人系统研究 |
通讯作者 | Shuo Wang |
作者单位 | 中国科学院自动化研究所 |
推荐引用方式 GB/T 7714 | Peng Hao,Tao Lu,Shaowei Cui,et al. SOZIL: Self-Optimal Zero-shot Imitation Learning, SOZIL: Self-Optimal Zero-shot Imitation Learning[J]. IEEE Trans on Cognitive and Developmental System, IEEE Trans on Cognitive and Developmental System,2021, 2021(1):1, 1. |
APA | Peng Hao,Tao Lu,Shaowei Cui,Junhang Wei,Shuo Wang,&Yinghao Cai.(2021).SOZIL: Self-Optimal Zero-shot Imitation Learning.IEEE Trans on Cognitive and Developmental System(1),1. |
MLA | Peng Hao,et al."SOZIL: Self-Optimal Zero-shot Imitation Learning".IEEE Trans on Cognitive and Developmental System .1(2021):1. |
入库方式: OAI收割
来源:自动化研究所
其他版本
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。