Controle inteligente do caminhar de robôs móveis simulados
AUTOR(ES)
Milton Roberto Heinen
DATA DE PUBLICAÇÃO
2007
RESUMO
The main goal of this dissertation is to propose, to test and to evaluate the use of Machine Learning (ML) techniques in the automatic con_guration of the gait control in legged robots. In order to achieve this goal, an extensive research about state-of-the-art techniques was accomplished and they are described in this work. This research allowed the development of the proposed model, called LegGen, which was implemented in a prototype. The proposed model allows the use of several different robot models with four, six or more paws. Besides that, the prototype allows also to study the robot s morphology evolution. The implemented prototype allows to accomplish experiments with autonomous legged robots, in a realistic three-dimensional virtual environment, through physics based simulations. The ODE (Open Dynamics Engine) software library was used in the physical simulation of rigid bodies and articulations, allowing to simulate forces acting in the articulations (actuators), gravity and collisions, among other physical properties of the objects inserted in the 3D environment. The implemented prototype also simulates sensors integrated in the robots, in order to control its state, stability and displacement. Several techniques of Machine Learning were studied to use in the control of the autonomous articulated legged robots. So, to control the articulations during the walk, four control strategies were proposed and implemented: (i) an automata based on angles tables; (ii) an automata based on positions tables; (iii) cyclic functions that describes the endpoints trajectory through a half-ellipse; and (iv) an Elman Arti_cial Neural Network. These control strategies have several parameters to con_gure, that are automatically obtained and optimized using Genetic Algorithms (GA). The GA were implemented into LegGen using the GALib software library. In order to improve the convergence and the gait control results, four _tness functions were proposed and validated using the prototype.From the execution of several experiments, it was possible to validate the proposed strategies of control, and it was also possible to accomplish a comparative study, presenting the advantages and disadvantages of each strategy. Besides that, several other experiments were accomplished comparing the different robot models, so that the most eficient model, according to a specific task, can be selected to be implemented in hardware.The experiments described in this work were validated through statistical methods and through graphical analysis of experimental data. A discussion about the experiments and posterior analysis of the obtained results, were also presented. Finally, the main contributions of this work are described, as well as the future work perspectives
ASSUNTO(S)
robótica móvel autônoma algoritmo genético ciencia da computacao intelligent control máquina genetic algorithms, artificial neural networks 3d virtual environments physical simulation articulated legged robots machine learning computação rede neural artificial autonomous robots aprendizagem
