Brief Functioning Description

The Robonauta Project provides a set of artificial agents and a robotic architecture that enhances the autonomous deployment of a guide robot into a museum. This task consists of three steps.

2D Mapping and Scene Tagging
A robot equipped with a laser range scanner, in our case Urbano, is launched into the scenario in which the guide robot will work. A person is in charge to guide the robot through the scenario. Using a person tracking algorithm the robot follow the person. When a place is to be tagged, for example a picture in the case of a museum, or a stand in a fair, the person tells the robot: "Tag this place as Las Meninas". The robot will store the coordinates of the place, and will tag it as Las Meninas.

After the whole scenario is mapped and tagged the person tells the robot "finish" in order to complete this task.

Ground and travesable areas detection

A major problem associated with traditional mobile robots is that their localization and mapping is based on 2D laser range scanners. A 2D sensor does not allow to detect obstacles that are above or bellow the sensor. This implies that there cannot be information about low or high obstacles, that could collide with the robot, and neither about holes, stairs, and so forth. In the Robonauta Project we approach this difficultie by providing the Nemo Robot with a tilted laser range scanner. This allows us to acquire 3D point clouds, identifying all obstacles in the scene. From the point cloud low, high and other non detected obstacles can be now detected and placed into the map. For this task, Nemo receives the map computed by Urbano and the trajectory that Urbano has followed. Nemo follows that trahectory, stopping at the convenient point and acquiring the 3D data, over-writting the Urbano map, and positioning the new obstacles detected. Doing this we have ready a traversability map, in which the safe areas to navigate are identified.

Robot Guiding
Finally, the traversability map, and the coordinates of targets are sent to the guide robot, in our case Doris. When Doris is started it begins wandering around the place, going randomply from one target to another. When it detects a person close to it, it stops and greets. Persons are free to request from the robot to be guided to a particular place, giving a voice command, as "show me Las Meninas". When the robot identifies such a command, it builds a safe path to the labeled point and goes there. Once it has arrived to the desired target it gives an explanation about the place, in this case, the picture Las Meninas.

3D Scene Reconstruction

Another task completed by the project is the 3D scene reconstruction, using photo-realistic information, in order to provide with frienly information to the user. Next figure shows an example of a 3D scene reconstruction of the CACSA Museum