Abstract
In this paper, two time shortning methods are proposed in path planning calculation of the autonomous robot vehicle using the neural network. The first method is to automatically set the relay point in order to avoid the local minimum in the generation of optimum route, and the second method is to give initial value of motion command input into the cascaded neural networks from the look-up table learned. These methods used jointly are examined for the effectiveness by applying to the path planning in various environments.
As the results, the effectiveness of the system which used two methods jointly at the start and immediately excluded the restraint of the relay point passage afterwards was clarified.
