University of Salamanca

P-NAV: Navigation system for a mobile robot

Department of Computer and Automatic science

    During last decade there has been investigated and experienced in the field of the navigation with mobile robots. In this period significant advances have taken place in the solutions to what has been a classic problem in Mobile Robotics that is the Goal-directed Navigation.

    In a similar way there has been introduced important concepts in the design of mobile robots that they have originated techniques of robots' construction for layers of growing complexity . In this last line the suppression architectures are included (subsumption  architectures) of the Prof. R. Brooks [Bro99] or the architectures based on behaviors (Behaviour Based Robotics) of the Prof. R. Arkin [Ark99]. On them they have been carried out numerous applications from those very you reactivate and directly connected with the environment to other more elaborated than they imply a hierarchization of performance levels. The basic virtue of these architectures is its applicability in problems of real time but
they present the inconvenience of not being able to have a representation of the environment beyond that that adquiren the sensors in each moment.

    In turn, the members of IEEE Yong K. Hwang and Narendra Ahuja have carried out studies on the navigation of mobile robots using like guide the value of the gradient of the potential in each point of the land to avoid obstacles..

    This project seeks to join two ideas. Of a part the techniques of exploration of the environment in real time by means of the calculation of potential fields. Of other, the tools of the systems based on behaviors and their implementation by means of technical of fuzzy control, as those used in environments like the Saphira developed in the SRI by the Prof. K. Konolige [Kor98].

    The purpose is to design and to implement a system that allows to navigate to a mobile robot using the information contained in the maps of the environment by the one that it navigates.  This is gotten starting from the calculation of the gradient of the potential in each moment of the place in which the robot is, thinking that as much the robot as the well-known obstacles are loads of the same sign and the destination point is a load of the contrary sign if we think of an electric potential. The robot will avoid collisions with the obstacles, he will carry out the local navigation and the planning of its movements in real time. The system will be implemented using the tools that Saphira has for the combination of behaviors.
 

 Back...