Start
•
Mecahnical Assembly •
GPS - Digital
Compass •
Proportional Servo •
Circuit Description •
Software •
Test Modes •
Crossing the Atlantic •
Sources
GPS
The
accuracy of this method for the basic, commercial
GPSs, is said to be within 100 m for the 99% of the
measurements collected in the same place during an
sufficient interval of time. (For more information
on GPS and its accuracy, see the series of articles
by Ingo Cyliax in Circuit Cellar 109–111.)
A
test was done with the GPS mounted on the Roboat.
A number of measurements were collected at the same
spot over a period of 50 min., and then plotted to
obtain the diagrams shown in Figures
3a and b. As you can see, the measurements actually
fluctuate in time with periods of roughly 5–10
min. and variations of ±50–60 m.
These
results confirm that the degree of accuracy for this
GPS is actually in the order of tens of meters. That
information was taken into account when planning the
course, so the Roboat never approached any obstacle
(island or a shoreline) at a distance closer than
100 m.
The
GPS device used in the projects is a Garmin GPS25-LVS
with the preamplified antenna GA27A. It sends a string
of ASCII data, according to the NMEA format, through
the serial output once every second, at a rate of
4800 bps (see Figure 4).
The
string is terminated by the ASCII codes "13"
(carriage return) and "10" (line feed).
Once a string is received, it is checked for reliability
by testing the checksum byte and by verifying that
the acquisition code is "A".
The
data that the program actually reads and uses are
the latitude and the longitude. Both latitude and
longitude are output by the GPS in the form of degrees,
minutes, and fractions of minutes, therefore the program
first converts them into "deci-milli-minutes
of degree," in order to obtain a single integer
number, without the troubles associated with degrees,
minutes, and fractions of minutes.
This
process is accomplished using the formulas:
latitude
= (dd × 60 + MM.mmnn) × 10000
longitude
= (Ddd × 60 + MM.mmnn) × 10000
In
assembler, the formulas are actually split into smaller
bits because the strings of ASCII data are first converted
into binary bytes (two characters at a time). Therefore,
the actual formulas are:
latitude
= ((dd × 60 + MM) × 100 + mm) × 100 + nn
longitude
= (((D × 100 + dd) × 60 + MM) × 100 + mm) × 100 +
nn
The
coordinates of the GPS are then compared to the coordinates
of the next course point, and the linear distances
to it are calculated expressed in meters, such as:
dX
= |longitude – course_point_long| × Xcf/10000
dY
= |latitude – course_point_lat| × Ycf/10000
Xcf
and Ycf are the conversion factors and should
be chosen according to the geographical area where
the Roboat is going to sail. The factors are computed
as:
Xcf
= meters/minutes of longitude ($512 in the test area
shown in Figure 2)
Ycf
= meters / minutes of latitude ($744 in the test area)
There
is no checking done for east or west longitude, or
for north or south latitude, so the rest of the Roboat
program works properly only with positive longitudes
and latitudes (as in most parts of Europe, Asia, and
North Africa).
