Precise GPS Sensing for High Performance Race Cars

Faculty Contacts: Jonathan How and John Hansman

Students: Chan Woo Park , Franz Busse , Nick Pohlman and Erica Peterson

The goal of this project is to design and develop a GPS sensing system that can
provide very high accuracy tracking (both position ~2 cm and attitude ~1 deg)
using Carrier-Phase Differential GPS (CDGPS).

This level of precision should enable:

• Real-time car / engine / tire performance monitoring
• Driver training and driving-line comparison with an expert.
• Post-drive analysis of track position and speed.
• Sideslip measurements.

Another exciting application would be to provide real-time car tracking during races
on television or the Web.

First test:

The first test was done in early August at Lime rock. We went there to take a three day
course at the Skip Barber Racing School. We drove the Formula Dodge cars

with the GPS antenna mounted on the top of the roll bar behind the drivers head
and the data from the GPS Receiverwas collected on a laptop located on the floor
under the driver's legs just in front of the seat.

As you might expect, the cockpit is pretty cramped.

	Photo of Prof. How driving around Limerock.
	Photo of Prof. Hansman getting into the racecar at Lime rock. The GPS antenna is mounted to the roll bar just behind him.
	Photo showing Chan and Franz at the reference station in the pit area

Unfortunately the results of this test were disappointing because the GPS receiver was
not setup correctly for the type of maneuvers experienced on the track.

Second test:

We recently modified the GPS receiver so that it works better in the highly dynamic
environment that will be experienced on the race track. We tested the receiver using Chan's car
on the roof of an MIT parking lot. The validation sensor was not available for these tests, but the
noise levels measured at the beginning and end of the run (car static) were very good.

	This plot shows the results from 4 trips around the parking lot at slow, medium, and high speed Note that there was no real attempt to follow the same course each time, so the tracks do not line up. Data points (collected at 5 Hz) shown by the marks on the plot.
	This plot shows the start/stop points in more detail. Note the significant change in scale. The scatter in the static results are clearly evident. Note that the starts were in the same place, but the stops varied depending on the speed.
	Plot focuses down in on the data at the start of each run. Note the significant change in scale from the previous plot. The car was placed as closely as possible back at the same spot to start each run. The start point is very close to the reference receiver antenna. Data was collected for several seconds before the run started.
	Plot focuses on the end of the high-speed (~40 mph) run Note the scale of the plot !
Slow Run	START END	Mean 0.0059m Mean 1.0293m	STD 0.0031m STD 0.0136m
Fast Run	START END	Mean 0.0067m Mean 0.4523m	STD 0.0037m STD 0.0063m

These tests are encouraging and more work continues to address the following key points:

• Maintaining GPS lock in the high-g environment in the racecar.
• Rapidly acquiring the biases after brief occulsions such as overpasses and bridges
• Validating the performance in the turns.

Other groups working on similar projects:

• FoxTrax by Sportvision
• Integrinautics
• SRI (Randy Galijan) - as described in a recent article in Road and Track Sept 2000.

Back