The goal of this project is to develop the technologies for a robot manipulator to perform autonomous object exploration of previously unseen objects and to iteratively adapt/refine/verify its own perception and manipulation skills. In particular, we want to demonstrate that the following are possible and practical:
- Unsupervised and safe exploration of a novel object.
- Automatic data collection, experiment labeling, and feature/parameter extraction.
- Iterative improvement and verification of manipulation skills for novel objects.
We want to study the processes of on-line data collection and skill testing. In some cases, a robot will have good priors from similar objects previously explored, and will need to verify them. In other cases, a robot will have to gather new data and update its manipulation behaviors.
[January-1-2018 to current]