Adriaenssens, Aurélien (2021) Testing For Generality Of A Proximal Policy Optimiser For Advanced Human Locomotion Beyond Walking. Bachelor's Thesis, Artificial Intelligence.
|
Text
Test_for_generality_of_a_PPO_policy_for_walking_behaviours_Final.pdf Download (3MB) | Preview |
|
![]() |
Text
toestemming.pdf Restricted to Registered users only Download (100kB) |
Abstract
Computer simulations have become a great aid in designing and testing new prosthesis before deployment in the real world. Being able to simulate for the human walking gait has shown time and cost saving to prosthesis development. With the creativity and dynamic environments of simulations, a multitude of environments can be tested upon. This paper focuses on determining whether a currently existing Proximal Policy Optimiser with Imitation Learning is able to be generalised to a variety of human locomotion beyond just walking. The environments tested on are: fast paced walking, ascending and descending a ramp, and ascending and descending stairs. Two musculoskeletal models are used; one healthy, and one transfemoral amputee. Both models are subjected to the environments. The forces exerted, joint angle, and rewards earned of the transfemoral amputee are compared to the healthy model to validate the research. Both models are able to make progress in the environments, exhibiting greater that 50% likeliness on the training data. It is suggested that generalising the Proximal Policy Optimiser for more advanced scenarios is indeed possible.
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: | 02 Sep 2021 11:33 |
Last Modified: | 15 Nov 2023 13:18 |
URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/26012 |
Actions (login required)
![]() |
View Item |