Lessons on the Reality-Gap: Iterations between Virtual and Real Robots
-
- Rosendo Andre
- Living Machine Laboratory, ShanghaiTech University
-
- Houseago Charlie
- Bio-inspired Robotics Laboratory, University of Cambridge
-
- Iida Fumiya
- Bio-inspired Robotics Laboratory, University of Cambridge
説明
Due to approximations between the virtual and real world, the knowledge transfer from simulations to robots is problematic. As physical conditions are prone to unknown and stochastic noise sources, the predictability reduces. We use data experiments from 100 real robots to tune the parameters of a simulation, and later used this tuned simulator to improve the design of the previous robots and find the optimum robot. We compare the simulated and observed behavior of this robot, and discuss our results.
収録刊行物
-
- 人工生命とロボットに関する国際会議予稿集
-
人工生命とロボットに関する国際会議予稿集 23 153-156, 2018-02-02
株式会社ALife Robotics
- Tweet
キーワード
詳細情報 詳細情報について
-
- CRID
- 1390845713049557120
-
- ISSN
- 21887829
-
- 本文言語コード
- en
-
- データソース種別
-
- JaLC
- Crossref
-
- 抄録ライセンスフラグ
- 使用不可