Mechanical design and MANFIS control of a leg for a new demining walking robot

Superior terrain adaptability has made legged robots candidate for many applications, inclusive demining. They offer attractive capabilities in terms of agility, obstacle avoidance and locomotion on soft ground. In order to build a walking robot for a demining application, a survey concerning the architecture of some selected walking project was made. This survey led us to choose the architecture of the new demining machine we intend to realise. A prototype leg has been built and the way it will be controlled is studied. The simulations of the MANFIS (Multiple Adaptive Neuro-Fuzzy Inference Systems) controller and the implementation of that control on the prototype leg are explained in this paper. MANFIS controllers have the properties of learning ability, parallel operation, structured knowledge representation, better integration with other control design methods. Those properties in a controller are necessary for making an intelligent machine, which has to respond to the environment that it explores.