Are GPS devices accurate enough to be used as evidence in accidents?
- Thread starter Cyclopathic
- Start date
Page may contain affiliate links. Please see terms for details.
classic33
Leg End Member
Less than 10 feet in 2001! Commercially available units.
http://www.spaceref.com/news/viewpr.html?pid=4919
Two handheld units, available now, can now go to less than three feet.
http://www.spaceref.com/news/viewpr.html?pid=4919
Two handheld units, available now, can now go to less than three feet.
benborp
Guru
- Location
- further from Penge
Just like PK99 I've had incidents where changes in speed, line and heart rate have been perfectly recorded with a scarily high fidelity (within inches rather than feet). There are however times when the GPS signal is so disrupted that my log struggles to place me within 50 yds of my actual location at any given time. The validity of GPS data would depend a lot on circumstance and also the steps taken to preserve evidence. I would consider handing over SD cards or devices at the scene if necessary. I wonder what protocols the police have in place for dealing with such evidence?
One instance where I would have fought to use GPS data if it had been available at the time was when I was collected by a car crossing from minor to minor across the South Circular. The owner was an utter ****. He lied through his teeth and knew exactly what to do to wriggle out of any responsibility. He didn't have to be consistent, just present a contrary argument to me to reduce the situation to being my word against his. Proving that I was actually there, riding in a certain direction and on a particular road would have been an immense help.
One instance where I would have fought to use GPS data if it had been available at the time was when I was collected by a car crossing from minor to minor across the South Circular. The owner was an utter ****. He lied through his teeth and knew exactly what to do to wriggle out of any responsibility. He didn't have to be consistent, just present a contrary argument to me to reduce the situation to being my word against his. Proving that I was actually there, riding in a certain direction and on a particular road would have been an immense help.
sidevalve
Über Member
Just one point here. To use such data you MUST accept that your every move can and will be tracked and could be used against you. It helps if BOTH parties are using a similar device and that both units are in 100% perfect conection [no possibility of any upload/download problems] at the very moment of the impact, [and it MUST be at the moment of the impact, even a 1 second gap can change a lot of things].
Finally sorry but I for one will fight to my last living gasp to ensure that NO ONE repeat NO ONE has the right to track or snoop on my exact movements in my own private life. For the pedants yes I know there are cameras etc and some [not my,] mobile phones can be tracked even when not in use but these are 'snapshots' not constant accurate snooping.
Finally sorry but I for one will fight to my last living gasp to ensure that NO ONE repeat NO ONE has the right to track or snoop on my exact movements in my own private life. For the pedants yes I know there are cameras etc and some [not my,] mobile phones can be tracked even when not in use but these are 'snapshots' not constant accurate snooping.
It's also worth actually looking at a log of a GPS. Only the most guilty of nerd would even think about editing enough of a file like this to make yourself appear slower or in a different place.
261,726 lines to edit over the 70mile ride I plucked that from.
Work like this is generally not done manually line by line, that's what scripts are for. It could be a billion lines long and it would still only take a matter of seconds.
Anyway, surprised this hasn't been posted in this thread yet, this guy had the makings of a hack, but stuffed it up :P
http://www.bbc.co.uk/news/uk-england-devon-11938711
IanSmithCSE
Guru
- Location
- Worcester, Worcestershire
Good morning,
I have a Nokia E5 phone with the Sports Tracker app, I appreciate that this is not a dedicated device but the same principles apply.
The app appears to take the position data quite seriously and doesn’t do clever things to average out the errors.
This means that sitting on my desk the phone has an average speed of close on 0.5mph when left untouched for a decent period, around an hour and the average speed flickers between 0 and about 0.4 mph peaking at almost 6 mph.
When used on a ride, the numbers are very, very close to the £8 trip computer with a wheel sensor, so these errors appear to be irrelevant in sensible usage. What they do show that there is sufficient variation between fixes to make it useless for assigning responsibility.
Bye
Ian
I have a Nokia E5 phone with the Sports Tracker app, I appreciate that this is not a dedicated device but the same principles apply.
The app appears to take the position data quite seriously and doesn’t do clever things to average out the errors.
This means that sitting on my desk the phone has an average speed of close on 0.5mph when left untouched for a decent period, around an hour and the average speed flickers between 0 and about 0.4 mph peaking at almost 6 mph.
When used on a ride, the numbers are very, very close to the £8 trip computer with a wheel sensor, so these errors appear to be irrelevant in sensible usage. What they do show that there is sufficient variation between fixes to make it useless for assigning responsibility.
Bye
Ian
Quite often the evidence of the accident is not the only information that is useful.
As with the insurance companies, it can be used to assess the driving as a whiole.
Proving that the driving prior to the accident was erratic, the driver was speeding, or had been accelerating and braking harshly can add context to the event
As with the insurance companies, it can be used to assess the driving as a whiole.
Proving that the driving prior to the accident was erratic, the driver was speeding, or had been accelerating and braking harshly can add context to the event
T.M.H.N.E.T
Rainbows aren't just for world champions
- Location
- Northern Ireland
He didn't hack anything though. I opened my logs in an open source version of Excel! and I could have changed anything I desired (not that I need to) All he changed was the date and he had 6 months to do it.
But then of course, how many people have access to or the know-how on scripting,then factor in type of accident. All of which will be similar but wholly different.
It probably has more to do with how many satellites you use. More satellites = more precise fix. Perhaps you need military-grade hardware to reach satellites further away than consumer stuff can talk to, or like you say, they limit consumer stuff so that the network isn't overloaded with everyone talking to 30 satellites at once.
classic33
Leg End Member
How many people listen to the same radio station at the same time or watch the same tv channel at the same time with any real loss of signal! Enviroment has a greater impact than the number of people receiving the same signal.
I was using the phrase 'hack' in the hacking something to make it work type, not trying to say he broke into a computer or something.
T.M.H.N.E.T
Rainbows aren't just for world champions
- Location
- Northern Ireland
Yes. I figured
swansonj
Guru
To get a fix at all, you need to see the signal from 4 satellites. Picking up further satellites after the first 4 gives an improvement in accuracy, because it allows you to average out the random errors from each individual satellite over a larger number of satellites. But it's only a marginal improvement in accuracy, because the random errors from each satellite aren't the biggest sources of error. Most of the time, most units will be detecting 7, 8 or more satellites- for many smartphones (but not Iphones owing to Apple's perversity) you can get apps that display the individual satellites.
The biggest source of error is common-mode errors (that affect the signals from all satellites) of which the biggest is the variable atmospheric conditions that the signals have to travel through. But because that affects the signals from the satellites, two different units at the same location should be roughly equally affected - they may both have an absolute error of 5 m or 10 m, but because that error is the same for both of them, their relative error is much smaller (that was the point being made by someone else upthread). They won't, however, have exactly the same absolute error unless they are identical units, because different manufacturers use different approaches for averaging the signals when there are more than 4 satellites, and different tricks to try to reduce errors.
To reduce the error, the commonest way is some sort of differential GPS. You use a base station of a fixed location, you look at what location the GPS is telling you that fixed point has, which tells you the current error, then you broadcast that error to all the other GPS units so they can subtract it. There are umpteen different commercial systems doing this - one used to be broadcast on the Classic FM carrier, there are some broadcast on dedicated frequencies, and several broadcast from satellites (not the GPS satellites, but usually a geostationary satellite like Astra). If you can access these signals (which usually requires paying a subscription) you can get cm accuracy. You can also use a local differential signal - set up your own fixed point in the corner of your field and broadcast your own correction signal to your farm or building site.
The ultimate accuracy as used by surveyors uses something called RTK, which measures the phase of the carrier signal instead of the modulation.
As someone said, the US turned off the "dither" (Selective Availability) they used to apply to civilian signals - from memory it was Bill Clinton who decided that. But the military still get better accuracy by accessing a different signal broadcast by the same satellites.
I'm not au fait with which, if any, of these systems sports GPS uses, I'm afraid. But it's inherent that altitude gains calculated by GPS are often seriously exaggerated. You have a sequence of fixes at some interval. Each one has a random error of say 1 m. How does the unit know whether the 1 m difference between this fix and the previous one is because of random error or because you've actually climbed 1 m? So they tend to add them up anyway with some degree of filtering to try to filter out the errors, but inevitably including some errors with the genuine changes, thus making the altitude change greater than it really was.
The biggest source of error is common-mode errors (that affect the signals from all satellites) of which the biggest is the variable atmospheric conditions that the signals have to travel through. But because that affects the signals from the satellites, two different units at the same location should be roughly equally affected - they may both have an absolute error of 5 m or 10 m, but because that error is the same for both of them, their relative error is much smaller (that was the point being made by someone else upthread). They won't, however, have exactly the same absolute error unless they are identical units, because different manufacturers use different approaches for averaging the signals when there are more than 4 satellites, and different tricks to try to reduce errors.
To reduce the error, the commonest way is some sort of differential GPS. You use a base station of a fixed location, you look at what location the GPS is telling you that fixed point has, which tells you the current error, then you broadcast that error to all the other GPS units so they can subtract it. There are umpteen different commercial systems doing this - one used to be broadcast on the Classic FM carrier, there are some broadcast on dedicated frequencies, and several broadcast from satellites (not the GPS satellites, but usually a geostationary satellite like Astra). If you can access these signals (which usually requires paying a subscription) you can get cm accuracy. You can also use a local differential signal - set up your own fixed point in the corner of your field and broadcast your own correction signal to your farm or building site.
The ultimate accuracy as used by surveyors uses something called RTK, which measures the phase of the carrier signal instead of the modulation.
As someone said, the US turned off the "dither" (Selective Availability) they used to apply to civilian signals - from memory it was Bill Clinton who decided that. But the military still get better accuracy by accessing a different signal broadcast by the same satellites.
I'm not au fait with which, if any, of these systems sports GPS uses, I'm afraid. But it's inherent that altitude gains calculated by GPS are often seriously exaggerated. You have a sequence of fixes at some interval. Each one has a random error of say 1 m. How does the unit know whether the 1 m difference between this fix and the previous one is because of random error or because you've actually climbed 1 m? So they tend to add them up anyway with some degree of filtering to try to filter out the errors, but inevitably including some errors with the genuine changes, thus making the altitude change greater than it really was.
Accuracy Values by Garmin Receivers
The declaration of the accuracy by Garmin GPS receivers often leads to confusion. What does it mean if the receiver states an accuracy of 4 m? This readout refers to the so-called 50 % CEP (Circular Error Probable). This means that 50 % of all measurements are within a radius of 4 m. On the other hand, 50 % of all measured positions are outside of this radius. Furthermore, 95 % of all measured positions are within a circle of twice this radius and 98.9 % of all positions are within a circle of 2.55 the radius. In the given example, nearly all positions are within circle with a radius of 10 m. The determined position is in the worst case accurate to 10 m.
Accuracy of position determination
Alien8
Senior Moment
- Location
- (A little bit north of) Cambridge
The reason GPS vertical (altitude) error is significantly greater than horizontal error is primarily a matter of geometry - you can't track satellites below the horizon.
