Evaluation of Techniques for Merging Information from Distributed Robots into a Shared World Model

Abstract: In the robotics domain today there are many uncertainties. Sensors fail to provide complete and reliable information of the world, a world that is dynamic with objects moving in and out of a robot’s perception. One way to mitigate these uncertainties is to share and merge information between robots. However, information sharing itself brings uncertainties and questions that must be addressed. What information is useful to share? How reliable is it? How often should it be shared? How should this information be integrated with the information already present at the receiver? This thesis is intended to deal with the merging of shared information from several distributed robots; information that is mainly useful for determining the real world positions of objects. We have identified some techniques for merging shared information and tested these in various different situations in an effort to determine the most appropriate one. Our tests show that sharing information can be very beneficial when objects are occluded from the view of one or more robots. The tests also show that information sharing is useful for decreasing object position uncertainty.