MOET: Manipulation of Objects for the Extraction of Text
Text is everywhere in the modern world and often contains important information. Robots which work in domestic environments, especially in 'Assisted Living' applications would greatly benefit from being able to find and understand text as humans do.
A key component of such a capability is 'active vision', in the sense that intelligent manipulation of an object or the environment can aid a text detection system by revealing text which was initially hidden or partially occluded.
The aim of the project is to build a robotic system which can progressively explore objects in order to detect and identify any text which may be on them.
Initially the focus is on rigid objects that can be picked up and manipulated by the robot. In the future, work will also be done to look at manipulating non-rigid objects such as bags of peas or folded paper, in order to increase the applicability of the system.
The robot used in our research is a BERT (Bristol and Elumotion Robotic Torso) upper body humanoid robot.
David Willetts MP sums up the aims of the MOET project during the recent BRL opening ceremony.
Work on manipulation strategies presented at the ICRA 2013 Interactive Perception Workshop
For publications relating to this research project please visit our Research Publications page and search by lead author.