2020 Fiscal Year Annual Research Report
Understand human tool-use in motion and contact for robotic tooling
| Project/Area Number |
18K18130
|
|---|
| Research Institution | NTT Communication Science Laboratories |
|---|
Principal Investigator |
高木 敦士 日本電信電話株式会社NTTコミュニケーション科学基礎研究所, 人間情報研究部, 特別研究員 (70802362)
|
|---|
| Project Period (FY) |
2018-04-01 – 2021-03-31
|
| Keywords | interaction control / motor learning / robotic tooling |
|---|
| Outline of Annual Research Achievements |
In FY2020, we succeeded at showing that humans learning a real or a virtual tooling task with force feedback exhibited the same learning patterns judged on a variety of metrics (peak force, arm's stiffness, tool velocity). Our results showed that skilled users could move and insert the tool in a single ballistic motion, suggesting that human motion planning optimizes the tool's motion and contact phases together, unlike most state-of-the-art algorithms which plan and control each phase separately. Based on these observations and insights from our earlier studies in this project, we have developed a robotic controller robust to feedback delays and sudden changes in the environment, in preparation for publication.
|
