Generalizable Imitation Learning Through Pre-Trained Representations
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2023
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915189961523200 |
|---|---|
| author | Chang, Wei-Di Hogan, Francois Fujimoto, Scott Meger, David Dudek, Gregory |
| author_facet | Chang, Wei-Di Hogan, Francois Fujimoto, Scott Meger, David Dudek, Gregory |
| contents | In this paper, we leverage self-supervised vision transformer models and their emergent semantic abilities to improve the generalization abilities of imitation learning policies. We introduce DVK, an imitation learning algorithm that leverages rich pre-trained Visual Transformer patch-level embeddings to obtain better generalization when learning through demonstrations. Our learner sees the world by clustering appearance features into groups associated with semantic concepts, forming stable keypoints that generalize across a wide range of appearance variations and object types. We demonstrate how this representation enables generalized behaviour by evaluating imitation learning across a diverse dataset of object manipulation tasks. To facilitate further study of generalization in Imitation Learning, all of our code for the method and evaluation, as well as the dataset, is made available. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2311_09350 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Generalizable Imitation Learning Through Pre-Trained Representations Chang, Wei-Di Hogan, Francois Fujimoto, Scott Meger, David Dudek, Gregory Robotics Artificial Intelligence In this paper, we leverage self-supervised vision transformer models and their emergent semantic abilities to improve the generalization abilities of imitation learning policies. We introduce DVK, an imitation learning algorithm that leverages rich pre-trained Visual Transformer patch-level embeddings to obtain better generalization when learning through demonstrations. Our learner sees the world by clustering appearance features into groups associated with semantic concepts, forming stable keypoints that generalize across a wide range of appearance variations and object types. We demonstrate how this representation enables generalized behaviour by evaluating imitation learning across a diverse dataset of object manipulation tasks. To facilitate further study of generalization in Imitation Learning, all of our code for the method and evaluation, as well as the dataset, is made available. |
| title | Generalizable Imitation Learning Through Pre-Trained Representations |
| topic | Robotics Artificial Intelligence |
| url | https://arxiv.org/abs/2311.09350 |