Geração de trajetorias para robos moveis autonomos via redes neurais artificiais
AUTOR(ES)
Jose Fabio Abreu de Andrade
DATA DE PUBLICAÇÃO
2001
RESUMO
The autonomous navigation of vehic1es has been an area of great interest for researchers for a long time, mainly in the area of artificial intelligence. One of the most interesting problems is that related to the trajectory generated for a vehic1e, which shall be guided in an environrnent, avoiding colliding with obstac1es in order to reach predefined positions. This thesis presents a methodology to generate trajectories using neural networks applied to mobile robot navigation. In a first approach, it has been used a single MLP neural network, with a 28-20-2 architecture, to solve this problem. However, this system presented a high computational effort for training the neural network, since the number of pattems increased due to the generalization of the environments. In a second approach this neural network has been improved to a hybrid system composed of logic operations and 36 MLP neural networks. The results of computer simulation presented graphically showed that the proposed methodology were efficient to guide the robot in unknown environments
ASSUNTO(S)
redes neurais (computação) detectores inteligencia artificial sistemas de veiculos auto-guiados robos moveis
ACESSO AO ARTIGO
http://libdigi.unicamp.br/document/?code=vtls000235867Documentos Relacionados
- Sistema de navegação para robôs móveis autônomos
- Ferramentas computacionais para robos moveis autonomos
- Sistema de controle híbrido para robôs móveis autônomos
- B-Spline : CAEP - algoritmos culturais para a geração de trajetórias B-Spline de robôs móveis
- Planejamento de trajetorias de um manipulador robotico usando redes neurais artificiais