Zhang, Jiayun (2025) Low-Cost Imitation Learning for Dual-Arm Robots: Leveraging a Single Human Demonstration. Master's Thesis / Essay, Artificial Intelligence.
|
Text
mAI2025JiayunZhang.pdf Download (19MB) | Preview |
|
![[img]](https://fse.studenttheses.ub.rug.nl/style/images/fileicons/text.png) |
Text
Toestemming.pdf Restricted to Registered users only Download (243kB) |
Abstract
Imitation learning through human demonstration has become the mainstream method for current robotic learning field, its impressive superior performance has been shown in a huge amount of research. However, the expensive, complex and time-consuming data collection remains one of the main challenges of imitation learning. In this project, we proposed a framework that can automatically extract the 3D coordinates of hand keypoints from human demonstration videos and convert them into the trajectory of the robot end-effector. Based on this, we further developed a data augmentation method that can generate multiple robot end-effector trajectories with generalization capabilities from a single human demonstration. Finally, we utilized large language models to evaluate the generalization of the augmented data through in-context trajectory prediction, verifying the effectiveness of our method in generating high-quality, generalizable training data.
| Item Type: | Thesis (Master's Thesis / Essay) |
|---|---|
| Supervisor name: | Mohades Kasaei, S.H. and Tziafas, G.T. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Master's Thesis / Essay |
| Language: | English |
| Date Deposited: | 22 Jul 2025 08:19 |
| Last Modified: | 22 Jul 2025 08:19 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/36404 |
Actions (login required)
 |
View Item |