Output details
13 - Electrical and Electronic Engineering, Metallurgy and Materials
University of Surrey
A Probabilistic Framework for 3D Visual Object Representation
This paper presents a novel statistical approach to learning 3D visual object representation for 6D pose estimation. This approach was an essential component of the PACOPLUS project and allowed a robot to discover new objects by autonomous manipulation and learning their shape over time. This research enables automonous object learning for robots and helps move robots from the lab to urban environments. This approach has been used in EU projects Xperience (http://www.xperience.org) and Intellact (http://www.intellact.eu/). Long term exploitation includes implementation for the deployment of service and care robotics, and further projects are in preparation with industrial partners.