Active visual search by a humanoid robot

Abstract

Object search is a very common task we perform each time we need an object. Humanoid robots are multipurpose platforms and will need to use generic tools to extend their capacities. It must thus be able to look for objects, to localize and use them. A search behavior would be a great improvement in humanoid autonomy and a step forward toward their rise outside laboratories. Before starting a search behavior, the robot needs a model of the desired object. This model could be provided by an external mechanism, but a humanoid has all the required abilities to build that model by its own. An undergoing project in our laboratory, called the "Treasure hunting" aim at integrating in a unique cycle, the model building of an unknown object, and the search for that object in an unknown environment. With such a combined skill, the robot may incrementally build a knowledge of its surrounding environment and the object it has to manipulate without any a-priori models. Latter the robot would be able to find and recognize that object. The time constraint is crucial, as a reasonable limit has to be set on the time an end user can wait the robot to achieve its mission. This paper will focus on the search behavior and we assume that the object model is already created. © 2008 Springer-Verlag Berlin Heidelberg.