Learning Sequential Force Interaction Skills
Learning skills from kinesthetic demonstrations is a promising way of minimizing the gap between human manipulation abilities and those of robots. We propose an approach to learn sequential force interaction skills from such demonstrations. The demonstrations are decomposed into a set of movement pr...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-06-01
|
| Series: | Robotics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2218-6581/9/2/45 |
| _version_ | 1797565055325175808 |
|---|---|
| author | Simon Manschitz Michael Gienger Jens Kober Jan Peters |
| author_facet | Simon Manschitz Michael Gienger Jens Kober Jan Peters |
| author_sort | Simon Manschitz |
| collection | DOAJ |
| description | Learning skills from kinesthetic demonstrations is a promising way of minimizing the gap between human manipulation abilities and those of robots. We propose an approach to learn sequential force interaction skills from such demonstrations. The demonstrations are decomposed into a set of movement primitives by inferring the underlying sequential structure of the task. The decomposition is based on a novel probability distribution which we call Directional Normal Distribution. The distribution allows infering the movement primitive’s composition, i.e., its coordinate frames, control variables and target coordinates from the demonstrations. In addition, it permits determining an appropriate number of movement primitives for a task via model selection. After finding the task’s composition, the system learns to sequence the resulting movement primitives in order to be able to reproduce the task on a real robot. We evaluate the approach on three different tasks, unscrewing a light bulb, box stacking and box flipping. All tasks are kinesthetically demonstrated and then reproduced on a Barrett WAM robot. |
| first_indexed | 2024-03-10T19:05:34Z |
| format | Article |
| id | doaj.art-e71141380f164b49832c53e6ecb003a8 |
| institution | Directory Open Access Journal |
| issn | 2218-6581 |
| language | English |
| last_indexed | 2024-03-10T19:05:34Z |
| publishDate | 2020-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Robotics |
| spelling | doaj.art-e71141380f164b49832c53e6ecb003a82023-11-20T04:09:50ZengMDPI AGRobotics2218-65812020-06-01924510.3390/robotics9020045Learning Sequential Force Interaction SkillsSimon Manschitz0Michael Gienger1Jens Kober2Jan Peters3Honda Research Institute Europe, 63073 Offenbach, GermanyHonda Research Institute Europe, 63073 Offenbach, GermanyCognitive Robotics Department, Delft University of Technology, 2628 CD Delft, The NetherlandsInstitute for Intelligent Autonomous Systems, Technische Universität Darmstadt, 64289 Darmstadt, GermanyLearning skills from kinesthetic demonstrations is a promising way of minimizing the gap between human manipulation abilities and those of robots. We propose an approach to learn sequential force interaction skills from such demonstrations. The demonstrations are decomposed into a set of movement primitives by inferring the underlying sequential structure of the task. The decomposition is based on a novel probability distribution which we call Directional Normal Distribution. The distribution allows infering the movement primitive’s composition, i.e., its coordinate frames, control variables and target coordinates from the demonstrations. In addition, it permits determining an appropriate number of movement primitives for a task via model selection. After finding the task’s composition, the system learns to sequence the resulting movement primitives in order to be able to reproduce the task on a real robot. We evaluate the approach on three different tasks, unscrewing a light bulb, box stacking and box flipping. All tasks are kinesthetically demonstrated and then reproduced on a Barrett WAM robot.https://www.mdpi.com/2218-6581/9/2/45human-robot interactionmotor skill learninglearning from demonstrationbehavioral cloning |
| spellingShingle | Simon Manschitz Michael Gienger Jens Kober Jan Peters Learning Sequential Force Interaction Skills Robotics human-robot interaction motor skill learning learning from demonstration behavioral cloning |
| title | Learning Sequential Force Interaction Skills |
| title_full | Learning Sequential Force Interaction Skills |
| title_fullStr | Learning Sequential Force Interaction Skills |
| title_full_unstemmed | Learning Sequential Force Interaction Skills |
| title_short | Learning Sequential Force Interaction Skills |
| title_sort | learning sequential force interaction skills |
| topic | human-robot interaction motor skill learning learning from demonstration behavioral cloning |
| url | https://www.mdpi.com/2218-6581/9/2/45 |
| work_keys_str_mv | AT simonmanschitz learningsequentialforceinteractionskills AT michaelgienger learningsequentialforceinteractionskills AT jenskober learningsequentialforceinteractionskills AT janpeters learningsequentialforceinteractionskills |