dc.description.abstract | In this growing technology modern times, mostly encountered all things that
are computerization oriented. Based on that, things that initially related to
dynamic system and needs mathematical modeling is considered very difficult to
be developed with existing methods manually. The large scope of dimensions on
decision making and many other complex things caused this dynamic system
meets various types of difficulty and needs a new control system which the linear
one was not able to handle that complex things.
One way to reach the idea of developing a new control system that can
facilitate human work and mathematical modeling is to build an ‘intelligence’
factor in the control system, which is Neural Network. This ‘intelligence’ system
will be applied to a control system that called Truck Backer-Upper, which is
required some calculation to know how to make a truck can move backwards to
park its truck to the loading dock from certain position in an area. This neural
network controlling system is chosen using Backpropagation with Binary Sigmoid
as its activation function and using feedforward phase on its implementation. The
goal of this controlling system is resulting the right steering angle in every step of
the moving truck from the first position until its target.
In implementation, the final position of the truck in the loading dock is
nearly perfect. The accuracy level from 70 data samples for each variable is x =
99,770% ; y = 99,061% ; dan φ = 99,986%. From these implementations, it is
proven that neural controller can be used for adjusting system. | en_US |