A Study on Agent-Based Box-Manipulation Animation Using Deep Reinforcement Learning

IF 0.5 4区 计算机科学 Q4 COMPUTER SCIENCE, INFORMATION SYSTEMS
Hsiang-Yu Yang, Chien-Chou Wong, Sai-Keung Wong
{"title":"A Study on Agent-Based Box-Manipulation Animation Using Deep Reinforcement Learning","authors":"Hsiang-Yu Yang, Chien-Chou Wong, Sai-Keung Wong","doi":"10.6688/JISE.20210537(3).0003","DOIUrl":null,"url":null,"abstract":"This paper focuses on push-manipulation in an agent-based animation. A policy is learned in a learning session in which an agent perceives its own internal state and the surrounding environment and determines its actions. In each time step, the agent performs an action. Then it receives a reward that is a combination of different types of reward terms, including forward progress, orientation progress, collision avoidance, and finish time. Based on the received reward, the policy is improved gradually. We develop a system that controls an agent to transport a box. We investigate the effects of each reward term and study the impacts of various inputs on the performance of the agent in environments with obstacles. The inputs include the number of rays for perceiving the environment, obstacle settings, and box sizes. We performed some experiments and analyzed our findings in details. The experiment results show that the behaviors of agents are affected by the reward terms and various inputs in certain aspects, such as the movement smoothness of the agents, wandering about the box, loss of orientation, sensitivity about collision avoidance, and pushing styles.","PeriodicalId":50177,"journal":{"name":"Journal of Information Science and Engineering","volume":"16 4 1","pages":"535-551"},"PeriodicalIF":0.5000,"publicationDate":"2021-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Information Science and Engineering","FirstCategoryId":"94","ListUrlMain":"https://doi.org/10.6688/JISE.20210537(3).0003","RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"COMPUTER SCIENCE, INFORMATION SYSTEMS","Score":null,"Total":0}
引用次数: 1

Abstract

This paper focuses on push-manipulation in an agent-based animation. A policy is learned in a learning session in which an agent perceives its own internal state and the surrounding environment and determines its actions. In each time step, the agent performs an action. Then it receives a reward that is a combination of different types of reward terms, including forward progress, orientation progress, collision avoidance, and finish time. Based on the received reward, the policy is improved gradually. We develop a system that controls an agent to transport a box. We investigate the effects of each reward term and study the impacts of various inputs on the performance of the agent in environments with obstacles. The inputs include the number of rays for perceiving the environment, obstacle settings, and box sizes. We performed some experiments and analyzed our findings in details. The experiment results show that the behaviors of agents are affected by the reward terms and various inputs in certain aspects, such as the movement smoothness of the agents, wandering about the box, loss of orientation, sensitivity about collision avoidance, and pushing styles.
基于深度强化学习的基于agent的盒子操作动画研究
本文主要研究基于智能体的动画中的推操作。策略是在一个学习会话中学习的,在这个学习会话中,智能体感知自己的内部状态和周围环境,并决定自己的行动。在每个时间步中,代理执行一个操作。然后它会收到由不同类型的奖励条件组合而成的奖励,包括前进进度、方向进度、避免碰撞和完成时间。根据收到的奖励,逐步完善政策。我们开发了一个系统来控制代理来运输箱子。我们研究了每个奖励期限的影响,并研究了各种输入对智能体在障碍物环境中的性能的影响。输入包括用于感知环境的光线数量、障碍设置和盒子大小。我们做了一些实验,并详细分析了我们的发现。实验结果表明,智能体的行为在某些方面受到奖励条件和各种输入的影响,如智能体的运动平滑性、在盒子周围的漫游、方向丧失、对避碰的敏感性和推方式。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Journal of Information Science and Engineering
Journal of Information Science and Engineering 工程技术-计算机:信息系统
CiteScore
2.00
自引率
0.00%
发文量
4
审稿时长
8 months
期刊介绍: The Journal of Information Science and Engineering is dedicated to the dissemination of information on computer science, computer engineering, and computer systems. This journal encourages articles on original research in the areas of computer hardware, software, man-machine interface, theory and applications. tutorial papers in the above-mentioned areas, and state-of-the-art papers on various aspects of computer systems and applications.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信