GPS Use at Patriot Hills  


The GPS position estimation is of interest for the deployment of a mobile robot and scientific sensors during a field expedition in Antarctica. It specifically provides ground truth for ground penetrating radar measurements and is a complementing sensor for mobile robot position estimation.

In January 1998, a differential GPS system was deployed at Patriot Hills, Antarctica, as part of a sensor validation robotics expedition. The mobile station was deployed in a sled, connected to a personal computer to record position and status data. The static station was deployed in a location close to the main camp. A radio link kept the two receivers connected to provide in-line differential corrections.

This page summarizes the experiments performed and draws inferences of the utility of the differential GPS system for autonomous Antarctic robots and ground penetrating radar sensing.

The GPS differential system configuration is described. After that, simple application example of the GPS is presented, showing the physical position of the tents and other elements in the camp. Plots and statistics of the availability of satellites are shown to characterize the reliability of the system. Finally, favorable conclusions are presented regarding the use of GPS for robotics operations in Antarctica.

System Configuration

The system has two main elements: a fixed and a mobile stations. Both stations are GPS receivers with respective antennas that can estimate position information with the usual +/-50 m. accuracy. They work together to provide higher accuracy by subtracting the position readings and using the fact that one of them is static. The following paragraphs describe each station, the data output and the storage process.

Fixed GPS Station

The fixed station GPS receiver was installed inside a plastic case together with a power supply and a radio modem. The case was isolated with foam and only wires would come out of the case. This wires connected the GPS receiver with the GPS antenna, and the radio modem with a radio antenna. A power cable also connected the fixed station power bus to a tent power outlet. Connected to AC power by a power cable buried in snow (power present only when generator on). This station is located about 15 meters away from the tent.

This configuration generated a warning in the GPS receiver every time the system was powered again. It is necessary to have a UPS or battery for permanent power for base station. That also allow for more integration time, to obtain a precise reference position for the fixed-mobile stations.

Another issue was the position of both antennas. The radio modem antennas have to be as high as possible from the snow to achieve a better range of trouble less communication. After having communications problems at a few meters distance between both stations. We raised the fixed station radio modem antenna to 2 meter high. That improved the communications to a few hundred meters. Unfortunately, it was not possible to perform modifications to the mobile station.

Mobile Station

The mobile station in part of the radar-sled. It consist of several electronic units. Those of relevance for the GPS system are the GPS receiver, the radio modem and the portable computer. The GPS receiver is connected to the GPS antenna, and serially to the radio modem and the portable computer.
The power for all the units comes from a car battery. This battery is connected to a 12VDC bus and power is distributed to all units. The initial power distribution did not take into account the voltage drop in the lines, and the voltage at the GPS receiver was lower than 11V when the battery had medium charge. This prevented the proper operation of the receiver, that would shut down itself. The problem was addressed by making the wires shorter.

The GPS receiver output was transmitted serially to the sled computer and recorded together with the information coming from the other sensors of the sled unit. The format is :
00 sequence number unique number in one single file of GPS data
01 msec milliseconds in the present week
02 week weeks after GPS reference year
03 sats number of satellites available
04 newpos indicates a new position was calculated
05 clockfix indicates that it got a new time update
06 horiz horizontal position information available
07 vert vertical position information available
08 weighted
09 overdet
10 ionosph
11 filt L1
12 differ differential corrections available
13 rtk Real time kinematics corrections available
14 fixed Fixed RTK corrections available (max accuracy)
15 widelane
16 static
17 null1
18 null2
19 null3
20 degNS Latitude degrees coordinate
21 minNS Latitude minutes coordinate
22 secNS Latitude seconds coordinate
23 degEW Longitude degrees coordinate
24 minEW Longitude minutes coordinate
25 secEW Longitude seconds coordinate
26 height Estimated height
27 deltaX Meters to the east from the base station
28 deltaY Meters to the north from the base station
29 deltaZ Meters over the base station

Application Example

To understand what kind of information the GPS provides, see the following partial view of the project camp at Patriot Hills. Here you see two small tents, and the big sleeping tent. One idea to map the positions of them is to move the mobile GPS platform (sled) around them, following a path similar to the blue dotted line.
The GPS position is recorded every second and the resulting plot of the information is like the following picture.
However, sometimes the communication link between the mobile and fixed station fails. In those cases, the position vector is zero. Those are "outliers" that distort the information. Code is developed to get rid of those outliers and obtain a smoother plot of the positions. See the next graph, that also has squares added to the positions of the tents.

Satellite Statistics

The number of satellites available at any time is important because it determines the quality of the GPS position information. The number of satellites needed to determine just horizontal information is 3. Four are required for determining vertical position information. For the Magellan system, a minimum of 4 common satellites received at the mobile station and the fixed station is necessary to provide the maximum precision advertised.

The following graphs show the percentage of time with 3 or 4 or more satellites. There is no record for less than 3 satellites because the receivers would not output any information. However, every time the system was turned on, with the antenna clear of obstructions, the system took at most one minute to start outputting data.

Total time considered: 7.04 hours.

The satellite availability shows that

Static Test

The static test consisted in leaving the sled static in one position, and recording the GPS data over a long period of time. The variations of the position over time give a measure of the possible error when GPS is used to mark the position of an object, for later inspection or retrieval.

The scenario is the discovery of an interesting object by the autonomous explorer. The object is tagged and its GPS position is recorded. This information is transmitted to a human explorer for later assessment of the discovery.

The next figure shows the three variables (X,Y,Z) as recorded over a period of ten minutes. There is information of more than two contiguous time, but this gives a very good idea of how the position changes according to the GPS.

The variation in both horizontal axis is less than +/-2 cm. The variation in the vertical axis is less then +/-10 cm. This is enough precision and accuracy to relocate objects of the size of a small rock.


The system gives excellent accuracy for locating ground penetrating radar soundings. By attaching GPS information to each sounding it is possible to build maps of subsurface data.

It also works fine providing position information for marking locations within centimeter accuracy. This is specially useful to mark the position of autonomous machine findings (rocks and meteorites) for later human inspection and retrieval.

The results obtained indicate that differential GPS is a useful tool for the upcoming robotics expeditions at Patriot Hills, Antarctica.

Back to Experiment Results.

Robotic Search for Antarctic Meteorites 1998
All material on this page is property of NASA and Carnegie Mellon University. Any image or text
taken from this site and incorporated into another document without consent violates the Copyright
Law of the United States and the Berne International Copyright Agreement.
Send comments, questions, or suggestions to Dimitrios Apostolopoulos.
This document prepared by Michael Wagner.