(9) 
One of the challenging problems in the application of adaptive 
control to automatic plotting machines is that parameters such as K 
in equation (20) vary at rates which are comparable to the rates of 
change of the error quantity AD . Much of the present theory of 
adaptive systems concerns systems in which parameters vary slowly 
compared with the variation of error quantities. It is not approp- 
riate to discuss details of this problem here. The study given above 
is intended simply to show that & problem exists. Electronic scan- 
ning techniques that make possible highly effective forms of adaptive 
control are essential to automatic machines of optimum performance, 
In addition to the limitations on system performance caused by 
variation of basic control parametres such as K in equation (20), 
serious limitations are imposed by imperfections such as backlash in 
mechanical linkages. Backlash not only introduces an indeterminacy 
in tracking but also requires that the value of X and hence the 
quiekness of error correction be reduced if acceptable tracking per- 
formance is to be retained. For this reason, and of course to reduce 
the magnitude of the mass M that must be driven, mechanical linkages 
should be as simple as possible, 
One other difficulty associated with the control systems of 
automatic plotters should be mentioned. This difficulty is caused 
by measurement noise. In Figure 2, the electrical signal e that 
is to give an indication of plotting error is shown as being noisy. 
In equation (13) of our discussion of the system of Figure (2), the 
electrical signal e is assumed to be noiseless. 
In practice, a more realistic representation for the signal e 
than that given by equation (13) is, 
= : + 
e height *noise 
error 
- K eS. AD * e 
(22) 
n 
Equation (22) simply states that in addition to the desired electrical 
voltage K,«S. AD , the electronic scanning equipment generates by 
its very nature a randomly fluctuating voltage e, . This randomly 
fluctuating noise voltage causes random fluctuation of the tracking 
point about the topographic track. 
Any improvements in the design of plotters that will allow 
measurement noise to be reduced relative to the desired information 
signals are important not only in improving the smoothness of mapped 
tracks but also in facilitating the extraction of the information 
necessary to design optimum adaptive control systems. Such systems 
will permit a high average plotting speed to be achieved over terrain 
and photographic fields of the widest variability. 
Conclusion: 
The problems of designing highly effective control systems for 
automatic plotting machines are difficult and fascinating. These 
problems are gystem problems that involve the combined characteristics