E.A. Puente

Neural traffic sign recognition for autonomous vehicles

A vision-based vehicle guidance system working in road environments has three main roles: road detection, sign recognition, and obstacles detection. Traffic signs provide very valuable information about the road in order to provide safer and easy driving environment. Traffic signs are designed to be easily recognized by human drivers mainly because their colors and shapes are very different from natural environments. The algorithm presented in this paper makes the best use of these features. The algorithm has two main parts: 1) for detection using the colors and shapes of the signs; and 2) for classification using a neural network.<<ETX>>

Robot assembly system for the construction process automation

This paper presents a robot assembly system for the construction industry based on an articulated robot placed over a mobile platform. The assembly process and the robot has been developed under the computer integrated construction concept. Its main task is the assembly of blocks for the erection of walls in industrial buildings.

Increasing Intelligence in Autonomous Wheelchairs

The ideas shown in this paper have been developed having in mind the main goal of designing a completely autonomous wheelchair prototype. The state of the art of mobile robotics is compared with the new trends in the field. The idea of autonomy made us focus our research in extracting the best from conventional Intelligent Mobile Robots techniques, looking towards the concept of Autonomous Mobile Systems (AMS). In order to clarify the body of the presentation, some practical examples, developed at our laboratory (DISAM-UPM), are included in parallel with the main discourse.

The ICa Approach to Intelligent Autonomous Systems

Our main interest is the identification of engineering methods for building intelligent machines for real -industrial type- applications. These machines are capable of operation in non-controlled conditions by means of advanced control capabilities, interacting with uncertain environments and performing complex work in collaboration with humans.

Intelligent process control: the CONEX architecture

The CONEX architecture for complex process control has been designed in order to cope with some control problem in the process industry. This architecture allows the integrated operation of several control layers, ranging from direct PID-based control to model based intelligent control. The architecture is organized following the principle of increasing precision with decreasing intelligence, but has a integration schema based on a multiresolutional model of the process. This is a distributed architecture based on a message passing schema between control objects. This objects have a loosely coupled operation, enhancing the reliability of the overall system. This architecture is oriented towards control in complex process plants, where the correct operation -security, quality, throughput-of the system requires the use of knowledge form operators and process engineers. Nowadays an application of this architecture is under development for a cement kiln facility in Spain.

World Modelling and Sensor Data Fusion in a Non Static Environment. Application to Mobile Robots

Abstract We describe a world modelling method able to integrate static and moving objects existent in dynamic environments. The static world is modelled by using an occupancy grid. The method is capable of modelling several moving objects. Whereas measurements belonging to actual targets are processed using a Kalman filter to yield optimum estimates, all other measurements are used to create or maintain multiple hypothesis corresponding to possible mobile objects. The viability of the method has been tested in a real mobile robot. Portions of this research has been performed under the EEC ESPRIT 2483 Panorama Project.

Trajectory tracking for a mobile robot-An application to contour following

Some control algorithms for the contour following guidance module of a mobile robot are described, and their performance is analyzed. Different approaches such as classical, fuzzy and neural control techniques have been considered in order to optimize and smooth the trajectory of the mobile robot. The module controls a virtual vehicle, by means of two parameters: velocity and curvature. The algorithms have been first simulated and then tested on the UPM mobile platform. The best results have been obtained with classical control and fuzzy control.<<ETX>>

A neural network supervisor for behavioral primitives of autonomous systems

The authors present a neural network implementation of a fusion supervisor of primitive behavior to execute more complex robot behavior. The neural network implementation is part of an architecture for the execution of mobile robot tasks, which is composed of several primitive behaviors, in a simultaneous or concurrent fashion. The architecture allows for learning to take place. At the execution level, it incorporates the experience gained in executing primitive behavior as well as the overall task. The neural network has been trained to supervise the relative contributions of the various primitive robot behaviors to execute a given task. The neural network implementation has been tested within OPMOR, a simulation environment for mobile robots, and several results are presented. The performance of the neural network is adequate.<<ETX>>

OPMOR: optimization of motion control algorithms for mobile robots

OPMOR is a hierarchical, interactive, window-based, graphical oriented software environment for the specification, simulation, performance evaluation, and optimization of motion control algorithms for mobile robots. The software environment allows the definition of the robot operating environment, the robot sensor, the robot geometry, and the specification, development, and testing of sensor-based mobile robot control software. A feature of OPMOR is the simulation of real sensors, real operating environments, and actual robots. Robot control software developed under OPMOR runs with minor modifications on two mobile robots existing at the Madrid Polytechnic University. The software environment has its own graphical user interface implemented in X-Windows. Structured, unstructured, static, and dynamic environments can be modeled with OPMOR. Although OPMOR can be used to simulate any motion control paradigm, the authors successfully used it to study reactive systems using behavioral control strategies. Many configurations for the operating environment have been tried using a wide variety of algorithms and implementations, and some results are presented.<<ETX>>

Learning behavioral control by reinforcement for an autonomous mobile robot

We present an implementation of a reinforcement learning algorithm through the use of a special neural network topology, the AHC (adaptive heuristic critic). The AHC constitutes a fusion supervisor of primitive behaviours in order to execute more complex robot behaviours as for example go to goal. This fusion supervisor is part of an architecture for the execution of mobile robot tasks which are composed of several primitive behaviours which act in a simultaneous or concurrent fashion. The architecture allows for learning to take place at the execution level, it incorporates the experience gained in executing primitive behaviours as well as the overall task. The implementation of the autonomous learning approach has been tested within OPMOR, a simulation environment for mobile robots and with our mobile platform UPM Robuter. Both simulated and real results are presented. The performance of the AHC neural network is adequate. Portions of this work have been implemented in the EEC ESPRIT 2483 PANORAMA Project.<<ETX>>