INTELLIGENT TECHNIQUES FOR COGNITIVE MOBILE ROBOTS

摘要

Abstract: Mobile and/or autonomous robots represent a widely approached research subject. The most common problems associated to all the types of mobile robots refers to the spatial orientation system and to the mathematical models. Additionally, many difficulties are generated taking into consideration practical aspects of implementation, but the technology evolution made possible to develop smaller sized robots and to provide them with greater autonomy.Most of the existent mobile robots are wheeled ones, easy to build, but not always easy to control. They are not at all purpose convenient, but could avoid obstacles, without being able to climb them. For an evolution in more complex configuration spaces, worm-like robots were also developed. Control strategies for such robots should have more or less embedded intelligence. Keywords : autonomy, mobile robots, intelligent control, artificial neural networks, fuzzy control 1. INTRODUCTION The evolution of mobile robots and their application in different fields – including the domestic area – imposes increasing the autonomy level of such devices. The operation an autonomous mobile robot in a real world unstructured environment requires taking into consideration multiple aspects. The control strategy must be selected in order to operate under conditions of imprecision and uncertainty (prior knowledge about the environment is, in general, incomplete, uncertain and approximate, perceptually acquired information is also typically noisy and incomplete). In these unpredictable evolutions of the environment the autonomy of mobile robots could be obtained by increasing the level of intelligence of the controller. In the structure of the control system new functions must be incorporated in order to give to the robots the possibility to search the goals by adapting on the unpredictable environment. The control strategy of the mobile robots must be based on reactive and goal-achieving procedures, which includes the perception, learning, planning and behavior generation phases. These new kinds of functions could be implemented by using intelligent techniques into the autonomous control system. The architecture of an autonomous control system includes the execution level, the coordination level and the strategic level. Each level in this architecture has more or less incorporated “intelligence”. The distribution of the intelligent tasks on different levels is realized taking into account the “IPDI PRINCIPLE” (ncreasing