Observation Planning for Object Search by a Mobile Robot with Uncertain Recognition

In order to handle complex tasks in an unknown environment, a robot has to build a map with both free space information and objects type and location. We present an active vision system which first detects candidate objects using global detection mechanism, and later identifies them by moving the robot closer and by using a local recognition mechanism. Having multiple candidates and uncertain algorithm outcomes, we cast the problem as a Markov Decision Process. We exhibit the modelization process, the capability of online solver to quickly find a good action, and finally the implementation on a real robot. This implementation consists of a set of Robot Technology Components (RT components) implementing each part of our method.