Recognition of Smart Objects by a Mobile Robot using SIFT-based Image Recognition and Wireless Communication.

Abstract

In this work, we focus on the problem of objectlocation and recognition by an autonomous mobile robot. In ourapproach, the robot does not have any prior knowledge aboutthe form and multiplicity of the objects. The robot, however, isequipped with an onboard camera and both objects and robot arecapable of exchanging data by using a common low-cost, low-ratewireless technology, namely a TmoteSky mote. The small storagememory of the mote is used to store a simple communicationprotocol and a description of the physical appearance of theobject, encoded by means of a set of Scale Invariant FeatureTransform (SIFT) descriptors. The mobile robot queries thesurrounding smart objects by sending a broadcast query packetthrough the wireless interface. The smart objects that receivesuch a query reply by sending their ID and a selection of theSIFTS that describe their appearance. When a subset of the SIFTdescriptors extracted by the current image of the robot’s cameramatches the SIFT descriptors received from a smart objects, therobot can locate the object in its current view and autonomouslynavigate towards the object, interacting with it.