Efficient Training of Generalizable Visuomotor Policies via Control-Aware Augmentation
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866912284427681792 |
|---|---|
| author | Zhao, Yinuo Wu, Kun Yi, Tianjiao Xu, Zhiyuan Ju, Xiaozhu Che, Zhengping Liu, Chi Harold Tang, Jian |
| author_facet | Zhao, Yinuo Wu, Kun Yi, Tianjiao Xu, Zhiyuan Ju, Xiaozhu Che, Zhengping Liu, Chi Harold Tang, Jian |
| contents | Improving generalization is one key challenge in embodied AI, where obtaining large-scale datasets across diverse scenarios is costly. Traditional weak augmentations, such as cropping and flipping, are insufficient for improving a model's performance in new environments. Existing data augmentation methods often disrupt task-relevant information in images, potentially degrading performance. To overcome these challenges, we introduce EAGLE, an efficient training framework for generalizable visuomotor policies that improves upon existing methods by (1) enhancing generalization by applying augmentation only to control-related regions identified through a self-supervised control-aware mask and (2) improving training stability and efficiency by distilling knowledge from an expert to a visuomotor student policy, which is then deployed to unseen environments without further fine-tuning. Comprehensive experiments on three domains, including the DMControl Generalization Benchmark, the enhanced Robot Manipulation Distraction Benchmark, and a long-sequential drawer-opening task, validate the effectiveness of our method. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_09258 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Efficient Training of Generalizable Visuomotor Policies via Control-Aware Augmentation Zhao, Yinuo Wu, Kun Yi, Tianjiao Xu, Zhiyuan Ju, Xiaozhu Che, Zhengping Liu, Chi Harold Tang, Jian Robotics Computer Vision and Pattern Recognition Improving generalization is one key challenge in embodied AI, where obtaining large-scale datasets across diverse scenarios is costly. Traditional weak augmentations, such as cropping and flipping, are insufficient for improving a model's performance in new environments. Existing data augmentation methods often disrupt task-relevant information in images, potentially degrading performance. To overcome these challenges, we introduce EAGLE, an efficient training framework for generalizable visuomotor policies that improves upon existing methods by (1) enhancing generalization by applying augmentation only to control-related regions identified through a self-supervised control-aware mask and (2) improving training stability and efficiency by distilling knowledge from an expert to a visuomotor student policy, which is then deployed to unseen environments without further fine-tuning. Comprehensive experiments on three domains, including the DMControl Generalization Benchmark, the enhanced Robot Manipulation Distraction Benchmark, and a long-sequential drawer-opening task, validate the effectiveness of our method. |
| title | Efficient Training of Generalizable Visuomotor Policies via Control-Aware Augmentation |
| topic | Robotics Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2401.09258 |