Ricardo García Rosa

Fuzzy logic strategies to control an autonomous mobile robot

ABSTRACT The present work describes some strategies for allowing an autonomous mobile robot to maneuver in an unknown environment. Such strategies are based in fuzzy control rules that are determined to correspond to the linguistic control rules of human operators strategies. They are transformed into algorithms to control the steering and the speed of a mobile robot. Perception of the environment is simulated by means of ultrasonic sensors.

PARALLEL INFERENCE ENGINE FOR FUZZY CONTROLLERS

This paper explains how to structure and to deal with the knowledge of any fuzzy control problem in order to build a parallel inference engine for an universal fuzzy controller. Essentially the description of one fuzzy control problem is constituted by the inputs, outputs, and rules. A system that receives this description, written in a formal language, and stores it in a data base adequate for the defuzzification process has been produced. The system creates links among the inputs and the outputs, according to the rules, and the procedures to deal with these elements in a suitable manner for computing the action of control. The computation procedure can be implemented sequentially or in parallel in any way: simulation, real-time controller, VLSI, etc. The granularity of the tasks and data depends on the characteristics of the problem and the available resources. We have distributed the system over a network of transputers.

Software tools to obtain a distributed universal fuzzy controller

The authors have implemented a distributed universal fuzzy controller capable of adapting itself to deal with a specific process. Due to its distributed nature the controller can run sequentially, in a processor, or in parallel, in a net of processors when the application requires it. In order to implement the distributed universal fuzzy controller the authors have developed a system with the following components: (1) a language to describe the control problem in terms of fuzzy inputs, outputs and rules, defined by a Pb-grammar; (2) a language to represent internally the control problem in terms of elementary fuzzy processes defined by a Pe-grammar; (3) a library of the elementary fuzzy processes, mentioned in (2); and (4) a set of software tools. Among the software tools there are an extractor, a loader and an ender procedure. The extractor identifies the elementary and independent processes in which the fuzzy controller can be divided. This extractor receives the Pb-description of the control problem and it provides the Ps-description. The loader receives the Ps description and puts the elementary processes into the physical processors and also it starts its execution. When the user wants to finish the control task, the ender turns off the set of processes. The distributed universal fuzzy controller has been implemented sequentially and in parallel on a network of transputers, its performance is being tested in an industrial application: the tuning of high frequency amplifiers.<<ETX>>

Robot path planning in the configuration space with automatic obstacle transformation

Abstract A method of solving the problem of finding the set of movements that allows a robot to go from a starting to an ending position while avoiding obstacles consists of using a path searcher in the configuration space. In order to achieve this, the complete space is divided into two disjoint subspaces. One of the two, the forbidden space, contains all the configurations leading to collisions; the other, the free space, contains all the configurations avoiding obstacles. The problem to be considered now is finding a path in the free subspace to maximize an optimal criterion. The generation of the free space is difficult to solve, and it is usually handled in an approximate way to permit an analytical treatment. Here we present a method of automatically generating the configuration set corresponding to an obstacle. The system is complemented by a path finder based on an algorithm of heuristic searching of solutions in state spaces.

Fuzzy Driving Strategies for Cars in Several Traffic Situations

Abstract The main idea underlying this paper is to cope with automatic pilots for cars by means of putting on board cars sensors and controllers. Among the available sensors DGPS are considered as the most convenient to estimate the position and orientation of the cars. The approach envisaged for controlling cars is to design elemental fuzzy controllers dealing with simple traffic situations. When complex traffic movements have to be carried out the fuzzy elemental controllers will be combined to obtain the global one. Simulations have been done taking into account that fuzzy controllers will run in fuzzy coprocessors.

Fuzzy Traffic Police for Autonomous Vehicles

The model and regulation of traffic systems are difficult tasks. Nevertheless Systems Theory and classic tools can be used to obtain the model and to regulate traffic respectively. A common way to organize traffic in a town area is to do a centralized control of the red lights based on queuing theory. Our approach is to distribute the control between a central system and on board car controllers that command every vehicle automatically according to the traffic rules. In order to do it we will implement fuzzy expert systems in computers augmented with fuzzy coprocessors and we will equip the vehicles with the necessary computers, instrumentation and communications.

A Qualitative System as a Frame to Control Unmanned Vehicles

It is known that the techniques bracketed under the topic of Soft Computing have a strong capability of learning and cognition joint to a good tolerance with uncertainty and imprecision. Due to these properties they can be applied successfully in Intelligent Vehicles Systems. In particular Fuzzy Logic is very adequate to build qualitative or linguistic models of many kinds of systems. The aim of this paper is to integrate in a qualitative model the vehicle operation and the driver behavior in such a way that an unmanned guiding system can be developed around it [2] [6].

Automatic Task Generator With Incomplete Information For A Robot Endowed With Sensors

We show an automatic task or plan generator which is able of finding the solution of a problem without knowing all the information about its current state. Therefore, for achieving the proposed goal, modifying actions over the real world, which will be carried out through an industrial robot, are not enough. Also searching information actions are essential. This is accomplished by sensorial devices. The chosen knowledge representation is a new production system, that is expanded by the parallel planning method and is modified by the addition of an optimizing level composed by heuristics or evaluation functions. The plan generator allows to solve problems that have interacting subgoals without resorting to backtracking, reordering partial plans and patches.