Abstract

A steering control of automated guided vehicles (AGVs) by neural networks was proposed. It was necessary to adjust the coefficients in the teaching signal according to the degree of learning. Thus, it was desired that they were adjusted automatically, then a steering control strategy of AGVs was proposed with the successive learning neural networks using the auto-tuning function (AT function). Since the change of traveling conditions, for example, a cornering radius and a traveling speed, are not considered, the steering control with the AT function is not practical. In this paper, a steering control strategy of AGVs is proposed with a successive learning neural networks using a real-time tuning function (RT function). The coefficients of the proposed RT function and the initial values of the teaching signal are discussed by the computer simulation. The right and left turnig experiments using the AGV built as a trial are performed and the validity of the RT function is discussed. The excellent traveling control of the AGV is obtained in the case that the traveling conditions are changed. Then the learning of the neural networks is almost terminated and after then the excellent traveling lasts. Thus, the proposed RT function is proved to be very available for the successive learning of neural networks.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call