NE S:
EE
EEE
i
I
ik
ie
UE
3
a
CET
Eee
SE
RSS
iom ER
ERE
PRE ee
SAGE TEE
= de
130 THE DESIGN OF PHOTOGRAMMETRIC PLOTTERS, HELAVA
completely known, or because a complete transformation could not be executed. Fre-
quently the instrumental or model coordinates are considered identical with X, Y, and Z.
An operator can hardly handle more than three motions simultaneously. Tradition-
ally, this has meant operating the three instrumental coordinates. This is a good and
obvious solution for a projection type plotter and, in fact, other solutions are hardly con-
ceivable. In the case of an analytical plotter, the situation is different. A choice can be
made as to the motions to be controlled by the operator and by the servomechanisms, and
the performance of the final instrument depends greatly upon this choice. This aspect
is, therefore, worthy of special attention.
The coordinates that have been mentioned above give us a large selection of possible
choices. All combinations, however, do not make sense to start with, and some others are
practically the same, so the selection is not as large as it might appear. In the following
paragraphs some possible combinations are studied. In this study we shall call the motions
to be controlled by the operator the input and the motions to be servo-controlled the output.
A. Input: xy, y4, Z. Output: (x,—x,), (yı—ys) (applied to x, y» respectively), X, Y.
This alternative is attractive in that it leads to a small-size basic measuring instru-
ment. The total measuring ranges will not be larger than the sums of the coordinates of
one photograph and respective parallaxes. The servoing problem is both easy and dif-
ficult: The quantities (v,—a,) and (y,—9s,) are small and slowly changing, easily handled
by the servomechanism; the final map coordinates X and Y are larger and fast changing,
thus difficult from that point of view.
The problem of servo controls may be alleviated in two ways, both at the expense of
completeness of the solution. Firstly, if the coordinates (x', y', 2’) are used instead of the
final map coordinates, the magnitudes of the critical outputs can be reduced by utilizing
the differences (x,—w"), (yy—y’) instead of X and Y respectively. This means that the
plotting results must be transferred to the map coordinate system by a separate proce-
dure, as in present practice. Secondly, the output quantities X and Y may be obtained as
a numerical record only. This would not require servoing of these large quantities and
would allow bridging, but no continuous plotting.
In this connection some other drawbacks of this solution should be mentioned. The
fact that the parallaxes are always corrected by the motions x, and y» complicates the
bridging procedure. It should also be emphasized that the original alternative A is not
favorable from the standpoint of process control. The two inputs X and Y are mot a part
of the control feedback loop. Therefore an error may be present in the output without
being observed.
B. Input: X, Y, 2. Output: 24, V1» or Yo
This solution is attractive due to its generality and symmetry. It has, however, two
drawbacks: The output quantities are rather large and fast changing, and the directions
of the input motions do not necessarily agree with the observed motions in the viewing
device. The latter drawback could be avoided by accepting the model coordinates, or the
instrumental coordinates as an input instead of X, Y, and Z. Then, however, additional
servoed outputs would be necessary to obtain a general solution, and these outputs would
be large and fast-changing, thus aggravating the problem of servomechanisms.
This approach has been employed, with some modifications, by Paul Rosenberg As-
sociates [4]. The most significant modification is that the input is in numerical form
without accompanying physical displacements. This makes it necessary to employ ad-
ditional servomechanisms.
C. Input: x, y, 2. Output: (y,—«), (y4—Vy), (*3—29), (/a—v).
This is the solution proposed earlier by the author. At the time of writing it seemed
3 8