Dexter scenario
In the Dexter Scenario we have developed robots that are concerned with different aspects of manipulation. We have developed three robot systems related to Dexter during the course of the project. The first of these Dexter systems learns to predict the effect of its actions on the motion of an object.
Dexter : Learning to predict the behaviour of manipulated objects from IRLab, University of Birmingham on Vimeo.
The second Dexter system estimates whether or not a grasp will be stable by combining estimates from vision and tactile sensing. Click on the image below to see a video describing how it works.
The Dexter system performs grasp stability estimation from vision and tactile information
The third Dexter system performs grasps on objects it hasn’t seen before. Click on the image below to see a video describing how it works.
Dexter is able to pick grasps for novel objects based on the transfer of parts