Modeling Objects as Aspect Transition Graphs to Support Manipulation
- Resource Type
- Authors
- Li Yang Ku; Roderic A. Grupen; Erik Learned-Miller
- Source
- Springer Proceedings in Advanced Robotics ISBN: 9783319609157
ISRR (2)
- Subject
- Infinite set
business.industry
Robonaut
ComputerApplications_COMPUTERSINOTHERSYSTEMS
Visual servoing
Conjunction (grammar)
Image (mathematics)
Task (project management)
Robot
Object model
Computer vision
Artificial intelligence
business
Mathematics
- Language
Designing robots that can model unstructured environments and act predictably in those environments is a challenging problem. In this article we address two issues in object modeling and manipulation. First, in image-based models, how can we choose a discrete set of canonical views from an infinite set of possible choices to support recognition and manipulation? Second, how can we make actions repeatable and predictable in unstructured environments? We propose an object model that handles both of these issues in a coherent way and introduce a novel image-based visual servoing algorithm that works in conjunction with the object model. We then demonstrate our object model and visual servoing algorithm on a tool grasping task on the Robonaut 2 simulator.