Research

Carlos Perez-Vidal, PhD Thesis (read on 4th July 2008)

Control de Robots Manipuladores Usando Informacion Visual:

Aplicacion a la Estimacion del Movimiento del Objeto

Control of robot manipulators using visual information:

Application to the object's motion estimation

Tecnologias Industriales

Universidad Miguel Hernandez

Departamento de Ingenieria de Sistemas Industriales

Visual servoing is a multidisciplinary research issue because it evolves the knowledge of different areas such as: robotics, control theory, electronics and computer vision. Visual servoing techniques are useful in many applications and are used to control different systems. However, there are a lot of unsolved problems that cause the non extending of visual servoing widely used in industry. One of the greatest limitations is the low speed at which these systems perform. This PhD Thesis contributes to the increase of the benefits of the visual control systems by accelerating their response.

In order to obtain this objective, the modelling and identification of a robot designed specifically for this type of applications is carried out. The developed robot is a cartesian robot of three degrees of freedom and this research presents a detailed model that allows the behaviour simulation before real experiments execution.

One of the contributions of this thesis is the design of a new prediction filter called Fuzzy Mix of Filters (FMF). This filter presents an improvement with respect to the classic filters in visual servoing schemes.

This document also presents a novel control scheme that considers the delay of the camera acquisition and processing. This scheme represents a considerable improvement with respect to the scheme used up till now.

This thesis has an important experimental component. Special attention to the implementation of FMF filter is given. The implementation of the control scheme and the acquisition/processing of the image is performed in an specific electronic device (an FPGA) in order to accelerate the computation time.