Kock, Robin (2022) Exploring Physics-Based Human Musculoskeletal Control using Deep Reinforcement Learning. Bachelor's Thesis, Artificial Intelligence.
|
Text
mAI_2022_RobinKock.pdf Download (1MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
toestemming.pdf Restricted to Registered users only Download (100kB) |
Abstract
This paper presents a Python library which provides reinforcement learning environments for bipedal musculoskeletal control. The environments can be configured with different rewards, observations and actuator controllers, while enabling easy cooperation between researchers. To demonstrate the library, three different rewards and three different observation spaces are tested. Our results suggest that the best performing observation space includes local body part positions but not the imitation data. Additionally, our best performing reward includes both a goal reward (reward for moving forward) as well as velocity-based and positional imitation reward.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Carloni, R. and Raveendranathan, V. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 01 Sep 2022 09:41 |
| Last Modified: | 12 Sep 2022 09:47 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/28645 |
Actions (login required)
 |
View Item |