Research Threads
Robotic Vision
- How can we build a virtual inventory model?
- How can we deal with a large object set?
- How can we deal with occlusion?
- How can we re-detection objects?
- How can we search for objects?
Manipulation strategies
- How can we select a good pick point?
- How can we minimize the risks of a retrieving an object?
- How can we include other sensor-modalities for grasp detection
- How does the content of the bin affect the extraction?
Success and Failure detection
- How can we predict the outcome of an action?
- How can we classify failure cases?
- How can we mitigate failures?
End-effector and sensor design
- What is a good gripper design?
- How can we include multi-modal sensory input?
Remote human assistance
- How can a human operator grasp an object?
- What is relevant?
- How can we minimize the time for a human to select a pick?
Continual learning and automated data collection
- How can we do the data collection?
