A Blog About Road Rallying

My license plate is 4ZEROS, and unless you are a rally enthusiast you either have no idea what it means or you think it stands for the 4 circles on the Audi it is mounted on. But to me, it means that vehicle is 4 getting the ZERO at the checkpoint or inmarker. A zero score is what every Road Rally team strives to get.

I remember running a monte style road rally, written by friend Ron, that was titled Zero Zeros. What did he mean by this. Luckily I figured it out. The trap of one leg was that the checkpoint was on a dead end road, yet the generals stated that missing a checkpoint was a 200 point penalty and that going through a checkpoint backwards was a 500 point penalty. Therefore is was an impossilbility to score a zero on the the leg. You eather got a 200 or a 500. About 1/2 up the dead end road to this infamous checkpoint, I realized the devilishness of my so called friend Ron, and what the title of the event meant. I immediately turned around and headed for the next checkpoint. By getting a 200 on that leg and not the 500 like everyone else, I won!

Some rallyist I know, are a ceratin class of people, and I know this article will impress the heck out of them. It is circa 2004, so the equipment has improved since, but the basics are the same. This article was published around June 2004 and was written by Kurt Dost

Now the meat of this article, another way to get the elusive zero score on a TSD Rally.

A GPS Solution for TSD Rallies

In order to match the rallymaster’s perfect time, rally teams use clocks that display fractions of minutes and not seconds, which will be the format for times shown in this article. Also, the perfect time will usually be truncated to 100ths of a minute, or two places, and these fractions of minutes are known as cents. TSD rally scores represent the deviation in cents from a perfect time. So, for our example, team A’s car arriving at the checkpoint at 12:06.72 would score a 0.

A TSD rally is not a race and arriving early is as “bad” as arriving late. For example, if a team arrives 2 hundredths of a minute early (12:06.70), then it receives a score of 2; if 33 hundredths of a minute late (12:07.05), a score of 33. A team cannot make up arriving late at a checkpoint by arriving early at the next, as each leg is scored independently. A rally consists of many legs, and the winner will be the team with the lowest cumulative score over all legs.

Measuring the Distance: The basic idea in rally computation can be stated in one of two ways: Given a starting time and the current time, at the CAST(s) assigned, have you traveled the correct distance? For example, if Team A was to start at 12:01.00 at a CAST of 32 and it is now 12:03.17 has the car traveled 1.158 miles? Or, given the mileage traveled at the CASTs assigned, does the current time match perfect time? For example, if Team A was to start at 12:01.00 at a CAST of 32 and the vehicle has traveled 1.158 miles, is the current time 12:03.17? Both these computations require that exact distance measurements are available to the rally navigator.

In calculating perfect times, the mileage is measured to thousandths of a mile. This is much more accurate than a standard automobile odometer is capable of showing – typically only displaying tenths of miles – but not actually of measuring. The rallymaster will use a precision odometer that reads out to three decimal places to obtain mileage between NRIs and then, using the same method as in the example, compute perfect time.

Most precision odometers work by using a magnet attached to an axle or wheel and a Hall effect sending unit, or transducer, which triggers upon detection of a strong magnetic field. Each turn of the tire will cause one pulse to be picked up by a computer, which will then simply add a value known as the calibration factor distance into the odometer, as seen in Figure 2. This calibration factor is based on the formula: pi 2 (Tire) Diameter 4 Circumference.

The calibration factor is usually expressed as a fraction of a mile, the number representing how far the vehicle travels for each revolution of the tire. What is needed now is the diameter of the tire that is on the vehicle to which the precision odometer is attached. This is found using the formula:

For a vehicle equipped with tires having a specification of P205/70R15 the formula yields 26.299 inches for the tire’s diameter and hence a circumference of 6.8851 feet. Thus, each pulse resulting from the wheel rolling along the road means the vehicle will have traveled 6.8851/5280 or .00130399 miles. This will be the calibration factor used for the precision odometer. Note that this is a theoretical value that won’t exactly match the rallymaster’s mileages and must be adjusted. All rallies allow for this adjustment in a segment of the rally known as the odometer calibration run to reflect real world conditions, such as tire load, tire inflation and road conditions.

In the 1980s, several hardware manufacturers – including Alfa, Terratrip, Zeron and Timewise – integrated precision odometers with a device to compute whether a TSD team was on time given the target CASTs. Some of these units are still manufactured and can cost up to $1,000.

The GPS Alternative John Fishbeck had competed in rallies in the 1960s when all that was available were mechanical calculators, such as a slide rule, along with paper and pencil, to compute in real time how close the driver was to maintaining perfect time. Now in the 21st century, having both a laptop computer and affordable portable GPS receivers available, he decided to develop a custom software-based rally computer, the SWRallyComp.

TSD rallies provide for several classes of competition. For instance, stock class allows a four-function calculator to be used. Equipped class allows any distance measuring equipment to be directly connected to any calculating device, which is the class the SWRallyComp is designed for. Figure 3 depicts the configuration of the system.

Several concerns were on the table in developing the SWRallyComp. One was whether a laptop running a Visual Basic program could effectively compute and display the necessary information in real time. Second, and more importantly, would the GPS fixes be accurate enough to compete with a precision odometer-driven rally computer.

The 12-channel receiver used in the SWRallyComp system includes a digital mapbase and display, graphical user interface, waypoint navigation, and the ability to use the Federal Aviation Administration’s Wide Area Augmentation system for DGPS positioning. The unit allows an external NMEA (National Marine Electronics Association standard) device, differential GPS (DGPS) beacon receiver, or portable computer to be connected to the GPS receiver through its serial port. Several formats are available for data transfer, one being the NMEA out setting. This setting results in the NMEA 0183 version 3.0 data being sent every two seconds to the serial port.

Corner Tracking Errors Developing a GPS-based distance-measurement alternative to a wheel turn sensor required close consideration of the road geometry of TSD rallies and the positioning function of a GPS receiver. A wheel-turn sensor and odometer use dead-reckoning techniques that link subsequent measurements directly to previous positions. GPS position fixes, however, are independent measurements that don’t necessarily account for a car’s travel along a road network with sharp corners.

For example, one of the concerns with obtaining GPS fixes every two seconds is demonstrated in Figure 4.

Assume the GPS fixes are taken at times t0, t1 and t2. The distance computed for Team A’s car for interval t0-1, segment t01t1, does not reflect the distance needed to negotiate the corner, segments t01tx and tx1t1. The computation for straight-line interval t1-2 is inherently more accurate. A smooth radius curve exhibits this same problem.

Increasing the GPS receiver’s positioning rate will reduce this corner tracking error. The GPS receiver used in the SWRallyComp can be set to use a proprietary format. This setting results in the PVT (position, velocity, time) Data Protocol being used to provide the PC with real-time PVT data, which is transmitted by the unit approximately once per second. Most club rallies are run at speeds from 25 to 45 mph, which means the vehicle travels anywhere from 36.67 feet to 66.00 feet every second. The one-second update yields a fix rate more in line with the needs of a car traveling 45 mph around curves in contrast, for instance, to a boat traveling 12 knots in a straight line on open seas.

Figure 5 provides an illustration of the corner tracking-error issue when negotiating a curve in a road. Team A is approaching a 90-degree turn. The radius r (radii OA and OB) of the turn will vary based on county road commission specifications. Assume the GPS obtains a fix every second, one at position A and one second later at position B. Depending on the speed of the vehicle, it will travel a distance along the arc AmB before the GPS will update its fix. The computed distance will then be chord AB.

Several similar arcs will comprise the 90 degree turn. The associated chord/arc discrepancies in distances are enumerated in Tables 2 and 3. (Radius figures are based loosely on the American Association of State Highway and Transportation Officials policy for expected vehicle speed through a curve for typical county aggregate roads on which rallies are run.)

The “overall difference” column shows the number of feet that are computed for the 90-degree turn short of what was actually traveled. The net effect of this for TSD rally purposes is that the GPS-driven rally computer will show it has traveled less distance than it actually has over the route. This will result in a team arriving earlier than they should at the checkpoints.

GPS Accuracy Accuracy can be expressed as the degree of conformance between a GPS receiver’s computed position and that of constant standards of reference, such as the World Geodetic System 1984 (WGS84) or regional map datums. Now that the practice of degrading the GPS signal – selective availability (SA) – is just a distant, unpleasant memory, a number of GPS receivers will provide absolute position accuracy within 15 meters. Using one of the real-time DGPS systems can reduce that error to within 3 meters, but this level of accuracy is not considered necessary for our application. This metric is usually termed absolute accuracy.

Another measurement can be termed relative accuracy, which does not reference an external standard. Relative accuracy is the ability to return to a visited position at a later time. Relative accuracy can also be a reflection of how static a fix remains. How do these accuracy measurements affect the use of a GPS for odometer computations?

The metric of absolute GPS accuracy can be thought of as follows: If I take a position fix at a landmark on earth, how close is the GPS fix to the actual coordinates of the landmark? For surveyors and pilots the answer to this question can be very important. If a pilot landing an airplane expects the end of Gerald R. Ford International Airport’s Runway 8R to be at 46° 52.59803′ N, 85° 32.53103′ W and the plane’s GPS receiver has a 10-meter error, then a problem could arise. This error is induced by several factors, most being of relatively long duration. For example, ionospheric conditions that delay the propagation of GPS signals from a satellite are noticeably affected as the Earth rotates with respect to the Sun but do not change greatly from minute to minute.

The relative metric is the more important to a rally GPS system. This measure of accuracy can be thought of as follows: Given a fix at position A, then moving a relatively small distance in a relatively small period of time to position B or, for that matter, remaining at position A, will the GPS receiver compute the difference, if any, correctly. Note that relative accuracy has nothing to do with absolute accuracy, as long as each fix is obtained under the same conditions. For purposes of distance measurement, a rally computer GPS fix does not need to be as accurate as a surveyed location, only that it is accurate relative to itself.

Each fix given by a GPS receiver has an estimated position error that can be visualized as a circle encompassing a percentage, ranging from 50 to 95 percent, of the fixes output. For the GPS receiver used in the SWRallyComp, this figure typically is between 5 and 10 meters. An important aspect of the relative error is that every position reported is not randomly scattered within this circle. In fact, for a non-moving GPS receiver each position output, though wandering, will be relatively close to the previous.

Figure 6 shows the progression of the position fixes at one second intervals for a stationary SWRallyComp unit. Using the data depicted in Figure 6, Figure 7 indicates the difference in feet each fix position is from the prior fix.

The average for the values in Figure 7 is .337 feet or 4.04 inches. Even though the differences are not zero and seem rather large for a stationary receiver, what effect will the relative error have on the SWRallyComp? First, this characteristic of fix positions moving though the rally team is stationary, for example at a stop sign, will result in the computer accumulating mileage as if moving, albeit very slowly. Secondly, for times when the team travels on roads heading north, south, east and west, will this relative movement cancel out?

The pattern in the progression of the fixes results in an inaccuracy in a direction. Assuming the direction bias or progression would be exhibited in the same direction regardless of the GPS movement, then some fix positions would result in more distance being calculated and some would result in less distance. For example, in Figure 6 several readings trend along longitude (radians) 1.4709644, with only small latitude changes.

If a vehicle were heading north at this time, then on average 4 inches would be added to the distance-traveled calculation in the one-second interval. However, if the vehicle were traveling south, then the distance computed would be short four inches. Given that each leg of a rally entails several directions of travel it would seem that these relative errors would, in effect, cancel out over the course of the event.

Functions of a Rally Navigator The navigator’s role is two-fold. First, he must assist the driver in staying on the prescribed route. This is done by using the NRIs provided by the rallymaster and making appropriate observations and decisions, such as noting what road to turn onto.

Second, in the non-equipped class, the navigator must provide the driver with a discrepancy in either distance or time from that of perfect distance or perfect time. Either value can be obtained using the CAST in effect and another known component of the TSD formula.

To obtain the mileage discrepancy, the navigator will take the CAST and multiply it by the time that they have been running at that CAST. For example, a CAST of 35 run for 1.50 minutes should place the car exactly .875 miles farther into the route from whence the CAST went into effect. The driver can then look at the odometer to see if they are over or under that distance. How much over or under is called the delta distance.

This delta distance is most often turned back into a time difference by dividing it by the CAST to give the driver a delta time. The delta time is given as a positive number to indicate the team is ahead of where they should be on the course, or as a negative number if behind and, hence, the driver should increase speed.

Of course, any general laptop computer can perform these functions much more quickly, and the navigator’s role in equipped class becomes one of route following and making sure CAST changes are put into effect at the correct NRI and at the precise landmark.

Functions of a Rally Computer All rally computers must be able to perform several basic functions, such as allowing input of leg starting times, mileages, and CASTs. It will then compute the needed information based on mileage data, input either manually or automatically.

_
The SWRallyComp, using mileage input derived from the GPS, allows the navigator to use function keys to effect pertinent actions at the appropriate time. Several of the function keys will bring up dialog boxes in which to enter data. We should note that the use of a laptop keyboard can be somewhat difficult in a car negotiating a corner at speed on a rutted gravel road. Figure 8 shows the main SWRallyComp form display.

The LCD driver module, shown in Figure 3, is mounted in front of the driver to provide pertinent information without having to look over to the laptop screen. The driver module can be configured to displayed specific fields. The following fields are on the SWRallyComp main screen:

Time Difference. The amount of time, in minutes and cents the team is ahead or behind perfect time. This value is also displayed on the driver module.

Actual Time. The time of day. The rallymaster and checkpoint workers will use time of day for leg start times. This time must be in synchronicity with the rallymaster’s clock. This is also known as current time.

Perfect Time. The time as computed using CAST’s and NRI’s given by the rallymaster. This value is computed based on the distance measuring equipment attached to the computer.

Miles to Tgt. This can be input if the NRI’s are mileaged. This mileage counts down to give the driver notification as to when to execute the next driver relevant action. This field can be displayed on the driver module.

Car Odometer/ Official. Mileage as accumulated on the rally leg or section. The Official mileage applies the Odometer CF (correction factor) value, as obtained on the ‘odometer calibration’ section of the rally.

CCAS/CAS. Change Average Speed To value in effect as indicated in the NRI’s. CCAS value applies the Odometer CF. This value is displayed on the driver module.

Next CCAS/Next CAS. Next CAST change. This will be the next value as indicated in the NRIs.

The navigator has the following function keys to effect TSD rally computations:

F1. Dialog Box: Set the leg start time, mileage and CAST. At the beginning of the rally each team is assigned a start time; thereafter, a rally worker at each checkpoint gives this to the team.

F2. Dialog box: Execute NRI. This makes an entry into a running log file, noting current time, perfect time and mileage. Useful for recovery purposes when the navigator discovers the team is off course.

F3. Dialog box: Set the next CAST. Allows the navigator to input the upcoming CAST as shown in the NRIs.

F4. Dialog box: Execute the CAST change. This will swap CAST and Next CAST.

F5. Dialog box: Execute a checkpoint. This makes an entry into a running log file.

F6. Dialog box: Adjust perfect time. Allows the input of PAUSE or GAIN values as shown in the NRIs. Pauses are input in cents; for example, a half-minute pause is input as 50.

F7. Dialog box: Allows navigator to modify the accumulated mileage (Official mileage).

F8. Dialog box: Allows navigator to modify the Odometer CF.

F9. Allows the navigator to PARK the computer when the car is stopped. This turns off any accumulation of mileage from the GPS. This is important, as the GPS will measure distance even if stationary. See Figure 6.

F10. Reverses the mileage accumulated. In effect subtracts the distance computed from the accumulated distance (Official mileage) in order to retrace portions of the route.

F11. Displays the running log in NOTEPAD. This is useful for recovery purposes (see function key F2).

Comparison of Systems Does the SWRallyComp GPS-based distance measuring paradigm compete with a hardware specific solution? We speculated that the errors introduced by GPS might make the system too imprecise. We also posited that the errors might cancel out, resulting in a fairly accurate system.

To test these hypotheses, on January 10, 2004, we used the SWRallyComp in the Son of Sno*Drift rally organized by members of the Detroit Region of the Sports Car Association of America (SCCA). Table 4 demonstrates that a rally run using the SWRallyComp with the 12-channel C/A-code receiver will allow a team to achieve zero point legs.

The leg values show whether a team is early (negative values), or late (positive values). The SWRallyComp has a propensity to calculate less distance than the rallymaster’s distance. This probably is because of the corner tracking error. Also the rallymaster may compute more distance for a leg as the result of tire slippage error that a drive train system introduces. In other words, “spinning out” will cause a wheel-driven odometer to measure more distance between locations than actually exists. Both errors result in the rallymaster ascribing more distance in the leg than the GPS-based system calculates. This may cause a rally team using the GPS-based system to arrive early at checkpoints.

Though the Fishbeck & Dost rally team does have some zero-hero scores, Leg 4’s error stands out. It would seem that the GPS relative errors cancel out in most cases. Leg 4’s distance measurement error resulting in the incorrect perfect time computation may have been caused by loss of satellite signal or multipath error.

Conclusion Stand-alone GPS fix information provides accurate enough positional change data to be used as an odometer input for a rally computer in equipped class, though not yet with the precision needed to achieve consistent perfect times. This seems to be a result of each GPS fix’s relative error being small and tending to cancel out as the vehicle’s direction changes over the course.

Further accuracy may be possible through updating fix positions faster than every second, which would decrease the corner tracking error. Mounting the external antenna outside of the vehicle, as opposed to an internal windshield mount, might increase signal strength. Finally, with a GPS modernization plan under way, nationwide real-time kinematic DGPS may yield positional data that will rival drivetrain-based precision odometers. c

Manufacturers The SWRallyComp developed by John Fishbeck used a GPS V receiver from GARMIN International, Olathe, Kansas. The laptop is a SlimNote Model P86 from Twinhead Corporation, Fremont, California, with an Intel Pentium II processor, 64 Megabytes of memory, with one serial port and one USB port used for attaching the GPS receiver and LCD Driver Module. The USB port/LCD Driver Module connection uses a USB/RS232 converter cable as the Driver Module’s interface is RS232. The operating system is MicroSoft Windows 2000 Professional.

Rate this:

2.5

Technorati Tags: Autombile Rally, car rallye, Fun with your car, GPS and Car Rally, GPS and Road Rally, rallies, rally, Rally Computer, rally timing equipment, rally with your car, road rallies, Road Rally Forum, TSD, TSD Rally, TSD rally and GPS