Comparing Skill Transfer Between Full Demonstrations and Segmented Sub-Tasks for Neural Dynamic Motion Primitives
Programming by demonstration has shown potential in reducing the technical barriers to teaching complex skills to robots. Dynamic motion primitives (DMPs) are an efficient method of learning trajectories from individual demonstrations using second-order dynamic equations. They can be expanded using...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-12-01
|
| Series: | Machines |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-1702/12/12/872 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1846103882676043776 |
|---|---|
| author | Geoffrey Hanks Gentiane Venture Yue Hu |
| author_facet | Geoffrey Hanks Gentiane Venture Yue Hu |
| author_sort | Geoffrey Hanks |
| collection | DOAJ |
| description | Programming by demonstration has shown potential in reducing the technical barriers to teaching complex skills to robots. Dynamic motion primitives (DMPs) are an efficient method of learning trajectories from individual demonstrations using second-order dynamic equations. They can be expanded using neural networks to learn longer and more complex skills. However, the length and complexity of a skill may come with trade-offs in terms of accuracy, the time required by experts, and task flexibility. This paper compares neural DMPs that learn from a full demonstration to those that learn from simpler sub-tasks for a pouring scenario in a framework that requires few demonstrations. While both methods were successful in completing the task, we find that the models trained using sub-tasks are more accurate and have more task flexibility but can require a larger investment from the human expert. |
| format | Article |
| id | doaj-art-bb8ee81b81844a209b29eaea224dd8d6 |
| institution | Kabale University |
| issn | 2075-1702 |
| language | English |
| publishDate | 2024-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Machines |
| spelling | doaj-art-bb8ee81b81844a209b29eaea224dd8d62024-12-27T14:37:02ZengMDPI AGMachines2075-17022024-12-01121287210.3390/machines12120872Comparing Skill Transfer Between Full Demonstrations and Segmented Sub-Tasks for Neural Dynamic Motion PrimitivesGeoffrey Hanks0Gentiane Venture1Yue Hu2Department of Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, ON N2L 3G1, CanadaFaculty of Engineering, The University of Tokyo, Tokyo 113-8654, JapanDepartment of Mechanical and Mechatronics Engineering, University of Waterloo, Waterloo, ON N2L 3G1, CanadaProgramming by demonstration has shown potential in reducing the technical barriers to teaching complex skills to robots. Dynamic motion primitives (DMPs) are an efficient method of learning trajectories from individual demonstrations using second-order dynamic equations. They can be expanded using neural networks to learn longer and more complex skills. However, the length and complexity of a skill may come with trade-offs in terms of accuracy, the time required by experts, and task flexibility. This paper compares neural DMPs that learn from a full demonstration to those that learn from simpler sub-tasks for a pouring scenario in a framework that requires few demonstrations. While both methods were successful in completing the task, we find that the models trained using sub-tasks are more accurate and have more task flexibility but can require a larger investment from the human expert.https://www.mdpi.com/2075-1702/12/12/872learning from demonstrationsrobot manipulatormachine learningneural dynamic motion primitives |
| spellingShingle | Geoffrey Hanks Gentiane Venture Yue Hu Comparing Skill Transfer Between Full Demonstrations and Segmented Sub-Tasks for Neural Dynamic Motion Primitives Machines learning from demonstrations robot manipulator machine learning neural dynamic motion primitives |
| title | Comparing Skill Transfer Between Full Demonstrations and Segmented Sub-Tasks for Neural Dynamic Motion Primitives |
| title_full | Comparing Skill Transfer Between Full Demonstrations and Segmented Sub-Tasks for Neural Dynamic Motion Primitives |
| title_fullStr | Comparing Skill Transfer Between Full Demonstrations and Segmented Sub-Tasks for Neural Dynamic Motion Primitives |
| title_full_unstemmed | Comparing Skill Transfer Between Full Demonstrations and Segmented Sub-Tasks for Neural Dynamic Motion Primitives |
| title_short | Comparing Skill Transfer Between Full Demonstrations and Segmented Sub-Tasks for Neural Dynamic Motion Primitives |
| title_sort | comparing skill transfer between full demonstrations and segmented sub tasks for neural dynamic motion primitives |
| topic | learning from demonstrations robot manipulator machine learning neural dynamic motion primitives |
| url | https://www.mdpi.com/2075-1702/12/12/872 |
| work_keys_str_mv | AT geoffreyhanks comparingskilltransferbetweenfulldemonstrationsandsegmentedsubtasksforneuraldynamicmotionprimitives AT gentianeventure comparingskilltransferbetweenfulldemonstrationsandsegmentedsubtasksforneuraldynamicmotionprimitives AT yuehu comparingskilltransferbetweenfulldemonstrationsandsegmentedsubtasksforneuraldynamicmotionprimitives |