|Accuracy of GPS||CFBlue|
Apr 21, 2002 7:52 AM
|I am pretty sure of the answer to this but
Yesterday was my longest ride of the season to date. 87 miles of the Sacramento Delta. The two bikes had two computers and a GPS.
I installed and set all the computers, doing the roll out rather than just pluging in the provided numbers. Not one of the three devices agreed with each other on total distance. The GPS recorded 87.1, and the computers in the lower 80's. We did take two short ferry rides across the river that the GPS recorded, but the computer didn't, but that would no way account for two or more miles different.
The Performance computer with cadence was closest to the GPS and my Cateye with the altimiter was way short.
So, my question is, should I trust that the mileage reported by the GPS is accurate?
|Pick one, don't obsess about it||SantaCruz|
Apr 21, 2002 8:04 AM
|re: Accuracy of GPS||davet|
Apr 21, 2002 9:08 AM
|The GPS is going to be the MOST accurate of the bunch, generally within a few yards. When you did the roll out for your computers you probably didn't take into consideration the diameter of the tires being less when loaded, therefore the difference in distances when ridden. And the difference between the two computers could be a difference in tire pressures, the loaded rolling diameter of the two hence being different.|
Apr 21, 2002 10:59 AM
|The GPS is going to be the worst of the three. Because it is only a handheld these are POS. They may give you accuracy to with in 10m at any given time. Now depending on the recording rate of the unit (time/ or calculated distance) you can see how the errors propogate to give such a large difference. I am not bash the technology(I am a surveyor and use it every other day) I am saying that the cycle computers are more accurate. The difference with these is in the way that they calculate roll out depending on brand depends on the formula(accuracy) used. Try riding a fixed known distance for calibration purposes.|
|Not 100% sure, but additionally||Lone Gunman|
Apr 21, 2002 11:14 AM
|the hand held GPS units have a built in inaccuracy of about 150 feet? or so. As I was learning about them for navigation during a flight school lesson, Federal govt decided that MFGs would build in this inaccuracy so that they could not be used for bomb targeting by civilians. At least this is what we were told and could not substitute a hand held GPS for instrument navigation.|
|i heard the same thing nm||ishmael|
Apr 21, 2002 11:46 AM
|since Clinton ended "selected availability"||cyclopathic|
Apr 21, 2002 3:43 PM
|most GPS have accuracy within 30'. This of cause includes most of "garden variety" 4 channel off-shelf build on Rockwell chipset. MIL spec units have much better accuracy, and top of the line over counter are within 10'.|
|Not 100% sure, but additionally||zeke|
Apr 21, 2002 8:49 PM
|i always get 12 to 15 feet accuracy on my handheld (garmin). that level of accuracy is available in japan, in the usa i think there is something called a WAAS system which gps owners can purchase, i think its an extra antenna???, which would increase the accuracy. also the US govt controls the satellites accuracy with something known as selective availabilty (SA). this is now in force because of the afghan thing, but if it were turned off/on? the accuracy of all gps units in the world would increase.
again, i will post an info site:
|Your bike computer if calibrated||MGS|
Apr 21, 2002 12:01 PM
|There are several problems with each system.
The bike computer must be correctly calibrated. Many people do a roll out measurement, but if you don't do it while riding the bike, i.e., compressing the tire, you will get a slightly inaccurate reading.
Additionally, it is very hard to get a perfect roll out as any wobble in the tire will give an inaccurate reading.
The GPS measures points and the draws a straight line between the points to give the mileage. If there is any short random wander between the points, the GPS will not meaure it. It assumes straight line travel.
Therefore, the computer is accurate, but your roll out may not be perfect.
Error built into civilian GPS units will not effect the distance measured as it has been recently turned off.
Apr 21, 2002 2:09 PM
|This was known as selective availability and was used to down grade the accuracy for civilian users. There are ways around this using different techniques RTK(real time kinematic) OTF(on the fly) and others. MGS was sort of right as I said the accuracy for distance measurement is also base on the timing interval for the points recorded. The shorter the interval the greater the accuracy. Also there are the issues associated with modelling of the ellipsoid. I could keep going but not point ranting. Basically you are interested in physical distance travelled not a projection distance so relie on the computer. The GPS is good for rough navigation but not distance measurement.|
|Then wouldn't the GPS show a shorter distance?||terry b|
Apr 21, 2002 4:44 PM
|If the GPS is connecting points (from satellite reads) via straight lines, then wouldn't the ultimate distance be less than even a reasonably calibrated computer? think about driving 10 miles of loopy ess curves. The computer is going to show the actual pavement distance (with some error to calibration) while the GPS would show a series of straight zigzags that would be a shorter total.
Anyway, I use an eTrex Vista, that has an accuracy of +/- 15 feet, not yards, meters or miles. I've used it on short bike rides and long car rides and it is extremely close to both my auto odometer and my bike computer. In fact, I regularly do a drive between here and Colorado Springs (357 miles) and the GPS usually agrees with the car by +/- 2 miles or so. The GPS, generally being longer, not shorter as you would expect if it's interpolating as you've decribed. On my last 1000 mile drive to Mexico, the GPS = the car.
The newer handhelds are extremely accurate whereas bike computers have lots of errors including the ones mentioned earlier (tire pressure, effect of persons weight) as well as missed magnet readings. In fact, a 30mm miscalibration on the distance they're talking about yields 1.3 miles error. In terms of GPS error, if it's constantly running a "track", then it's adding up all the extra feet you get from milling around that the bike computer would not see.
If my life depended on it though I'd bet it on a GPS over a 30 dollar bike computer any day.
|good counterargument (NM)||zeke|
Apr 21, 2002 8:52 PM
Apr 21, 2002 9:34 PM
|What you have is error propagation. If you travel a distance of say 10 miles(just for a you americans, which I think is 95% of the board) and lets say that you take a position every 1/2 mile. You are going to 20 points. Next lets say that the accuracy is +/- 15 feet. Now I stop because the next piece involves what is know as a least squares adjustment. And it propagates the errors through the network(your route). However that is what i do at work ans can't be naffed doing at home. It is possible that you can have less, equal to, or more than, the trip distance on the bike computer. This is dependent on time of day and satellite configuration etc. Also another factor is the unit being used. What I find interesting is the way that companies try to sell hand helds. If you look at the specs that dance right around the accuracy issues. Where as I know for a fact my work gear will give me 1-2cm RMS in postion and height. I don't mean to rubbish hand helds they just aren't as good as everyone thinks.|
Apr 21, 2002 3:33 PM
|I believe the computer would be more accurate. GPS signals come is intervals. So if you were not going straight (you wheels hits something and go awry for a fraction of a sec.), GPS will not take that into account. Meanwhile your cyclo-computer uses arithmetics to calculate distance. Multipling the circumference of your wheel by the number of revolutions.
However, the circumference of your wheel is a approximate number, so there is so error there. But even the military GPS have about a 6 yard error (I think).
Looking at the difference in your distance, it seems like it's very minimal. Say if the lower-80s distance you are referring to is 82 and the GPS distance is 87 and the actually distance is 84.5 the percentage of error is only about 3%. In statistics or science this would be considered a good approximation.
Plus, the Timex GPS watch is mainly geared toward runners. They can't just multiply stride length by stride number because the stride length will be all different.
I hope this helps.
|For Distance Travel Go with the Computer||SingleThreaded|
Apr 21, 2002 6:52 PM
|A typical GPS with S/A (Selective Availability) turned off will give you 10m positional accuracy (95% of the time). In most instances you can expect 5m. However, this performance is degraded when the aerial view of the satellites is obstructed by trees, buildings, ferry ceilings and big bike helmets. The probability of flyers (bad data points) is much higher on the GPS than the bike computer. These jogs in the GPS tracks, along with the 5-10 meter jitter with each fix only serve to increase the error in the distance recorded with every measurement.
Regarding the computers, I would agree with the previous post that a 30 mm error measurement would provide a 1.2 - 1.3 mile error over this distance. But this error, while realistic, approaches a poorly calibrated computer. The jitter of a wobbly wheel's path on the road is the actual distance traveled. Calibrate the computer over a known distance and reduce the error.
If I was betting my life on distance traveled, I would go with the properly calibrated computer. If I was betting on one of these devices saving my life, I would go with the GPS.
|re: Accuracy of bike computer.||Chen2|
Apr 21, 2002 7:04 PM
|Most of my rides are in areas with section line roads that are surveyed exactly one mile apart. I calibrate my bike computers to the centers of the intersections, haven't had much luck with the rollout technique. I can dial in my computeres to within 0.025 miles accuracy. This surprised me, I thought the natural lateral movement on the bike would throw off the accuracy more. I checked topographic maps to confirm that the intersections did not include any survey corrections. I adjust the wheel circumference calibration until I get the correct distance.
And bike computers that accept the wheel circumference in mm instead of cm are a magnitude more accurate.
|re: Accuracy of GPS||zeke|
Apr 21, 2002 8:43 PM
|it depends on a few factors, eg, does your gps unit have a setting for more or fewer points acquired per unit time? if so, of couse, more is more accurate; next the unit will be more accurate the less obstructed the space is between the unit and satellites, eg under trees is less accurate than riding on flat plains; also the older units have only 8 channels, while the newer have 12 and are more accurate.
you report about a 10% difference between the gps and computers, thats a bit much, are you sure the computers are accurate? have you tried the route by car and compared it to the odometer reading?
please check the attached url, they will be able to answer any questions you may have.
|re: Accuracy of GPS||zeke|
Apr 21, 2002 8:44 PM
|hmm, the url didnt appear in my message.
i will paste it here directly.
|re: Accuracy of GPS||jim hubbard|
Apr 21, 2002 9:39 PM
|the channels that manufacturers use refer to the number of satellies able to be tracked at anytime. Survey quality has always been 10+. As for the number of satellites being tracked at anytime 5 is the absolute minimum for a solution. Any more than this provides greater levels of redundancy in the solution. I can go through my old university notes if you require future information.|| |