Panoramic Representation of Scenes for Route Understanding

Building I This work tackles route understanding in robot navigation. The strategies employed are route description from experience and route recognition by visual information. In description phase, a new representation of scenes along a route termed Panoramic Representation is proposed. It is obtained by scanning scenes sideways along the route, which provides rich information such as a 2D projection of scenes called Panoramic View, a pathoriented 2(1/2)D sketch, and a path description. The continuous Panoramic View is more efficient in processing than integrating discrete views into a c o m p l q route model. In recognition phase, the robot matches the Panoramic Representation from incoming images with that memorized in the previous scan so that it can locate and orient itself in autonomous navigation. The advantage of wide field of Panoramic Views brings reliable scene recognition. Fig.1 A mobile robot continuously views scenes along a route by a camera. I t autonomously builds the model along the route, which guides the navigation alone the same route. A route shown in this figure is used for the experiments. The key step to bridge local episodic scenes along a route to a global semantic description for route understanding is, perhaps, encoding of route views into an easy-to-access presentation. rn this paper, a novel representation of scenes, called panoramic ~ ~ ~ ~ ~ ~ ~ ~ t ~ ~ i ~ ~ , is proposed, for memorizing and retrieving the information acquired in the trial move. This representation provides the essential information of scenes, such as 2D projection, a path-oriented 2(1/2)D &etch, and a path description. It can be used not only as an intermediate representation as M~~~ propose~[io], from which a more abstract and symbolic representation of the route can be built, but also as a description referred directly to in route recognition. In order to represent a wide field of view that contains global information with a small amount of data, we use Dynamic ~rojection[11,1zI which Scans Scenes through a moving vertical slit. We call it Panoramic View (PV) which has the following two types: Local Panoramic View (Lpv): Projection of around a stationary view point, memory of a place used for determining location and orientation of the robot. Route Panoramic View or simply Panoramic View: Projection of side-views along a route, memory of the route used for locating the robot. The 2(1/2)D sketch described by the image velocity of each feature, is extracted by measuring the time delay of features in appearing in a pair of parallel slits in the image. In route recognition phase, the robot recalls, the route memorized in its trial move to identify its location and orientation. We, therefore, need to match two panoramie representations obtained from different moves along the same route, assuming the road-following process keeps the path within the roads. We also match two local panoramic views at positions close