机构

• 长春光学精密机械与物... [1]
• 自动化研究所 [1]

采集方式

• OAI收割 [2]

内容类型

• 会议论文 [1]
• 期刊论文 [1]

发表日期

• 2020 [1]
• 2011 [1]

学科主题

筛选

浏览/检索结果: 共2条，第1-2条 帮助 条数/页： 5101520253035404550556065707580859095100 排序方式： 请选择题名升序题名降序提交时间升序提交时间降序作者升序作者降序发表日期升序发表日期降序 Real-Sim-Real Transfer for Real-World Robot Control Policy Learning with Deep Reinforcement Learning 期刊论文  OAI收割 APPLIED SCIENCES-BASEL, 2020, 卷号: 10, 期号: 5, 页码: 16 作者:  Liu, Naijun;  Cai, Yinghao;  Lu, Tao;  Wang, Rui;  Wang, Shuo   |  收藏  |  浏览/下载：15/0  |  提交时间：2020/06/02 robot  policy learning  reality gap  simulated environment  deep reinforcement learning   Reality Sim: A realistic environment for robot simulation platform of humanoid robot (EI CONFERENCE) 会议论文  OAI收割 5th International Conference on Automation, Robotics and Applications, ICARA 2011, December 6, 2011 - December 8, 2011, Wellington, New zealand 作者:  Fu Y. 收藏  |  浏览/下载：35/0  |  提交时间：2013/03/25 As a virtual training  testing and evaluating environment  simulation platform becomes a significant component in Soccer Robot project. Nevertheless  the simulated environment in a simulation platform usually has a big gap with the realistic world. In order to solve this issue  we demonstrate a more realistic simulation system which is called Reality Sim with numerous real images. By this system  the computer vision code could be easily tested on simulation platform. For this purpose  previously  an image database with a large quantity of images recorded by camera pose is built. Furthermore  if the camera pose of an image is not included in the database  an interpolation algorithm is used to reconstruct a brand-new realistic image of that pose such that a realistic image could be provided on every robot camera pose. Our results show this system effectively simulates a more realistic environment for simulation platform. 2011 IEEE.