Abstract
In critical human/robotic interactions such as, e.g., teleoperation by a disabled master or with insufficient bandwidth, it is highly desirable to have semi-autonomous robotic artifacts interact with a human being. Semi-autonomous grasping, for instance, consists of having a smart slave able to guess the master’s intentions and initiating a grasping sequence whenever the master wants to grasp an object
in the slave’s workspace. In this paper we investigate the possibility of building such an intelligent robotic artifact by training a machine learning system on data gathered from several human subjects while trying to grasp objects in a
teleoperation setup. In particular, we investigate the usefulness of gaze tracking in such a scenario. The resulting system must be light enough to be usable on-line and flexible enough to adapt to different masters, e.g., elderly and/or slow. The
outcome of the experiment is that such a system, based upon Support Vector Machines, meets all the requirements, being (a) highly accurate, (b) compact and fast, and (c) largely unaffected by the subjects’ diversity. It is also clearly shown
that gaze tracking significantly improves both the accuracy and compactness of the obtained models, if compared with the use of the hand position alone. The system can be trained with something like 3.5 minutes of human data in the worst case.task is neutral.