Design and operation of multiple trackless automatically guided vehicle systems

This paper examines the use of perception systems which can also be used to obtain images of the environment. This technique may be tedious if on a regular basis temporary obstacles are placed along the path of a vehicle. Therefore it is important to determine the path for the vehicle using an off-line technique. Computer aided drafting packages offer an interesting solution to this problem. It is possible to generate path data directly from the output format of a drawing file and use this to generate suitable training data for the vehicle off-line. The neural network weights can then be determined and downloaded to the vehicle. Finally the paper examines multi-vehicle environments, which will become more commonplace. Both track following and perception based vehicles also suffer the fact that they must avoid other vehicles and use complex proximity sensor systems in order to avoid each other. Using elementary communications and data generated from a CAD system it is possible to develop anti-collision algorithms which inform vehicles when to start and stop