Average speed cameras on A-roads - how do they work?

facade

Okell said:

 <snip>
The first camera records your car when it goes past whatever part of the road the camera is aimed at.  The same with the second camera.

While the aiming point on the road must be some distance from the camera lens (I suppose it could reasonably be between perhaps 20m and 50m from the camera) it is a fixed point and won't move.  
<snip>

Since I have tried to research this, I am surprised that I can't find an answer as to what the fixed point is, where it is in relation to the camera and how they ensure that there is a measured distance between the points when they don't place a marker on the road, which would seem to be the logical thing to do, and then aim the camera at it.

It shouldn't be a secret, as knowing the answer won't give to a way to beat the cameras.

Grey_Critic

Cameras do not record/measure speed. Camera A takes a picture when you pass the focal point and records the registration number of the vehicle.and the time. Camera B does the same. The system then does a calculation. based on the time taken for the vehicle to travel between the two or more points. There are invariably multiple cameras so it is basically contuosly recoding/checking.. If for example it should take 180 seconds at a posted limit and the vehicle does it in less time then you have exceeded the posted limit and the ticket is in the post. - It is AVERAGE so is not just two cameras.

prowla

The ones I've seen are aimed at the rear of the cars and I haven't particularly noticed any significant variation in the angle of the cameras.

The simplest way to think about the calculation is to suppose the cameras A and B were sited a mile apart and the speed limit was 60, then if B saw you less than a minute after A, then you must've been going faster than a mile-a-minute, or 60 mph.

I would guess that they factor enough wiggle-room into the calculation to cover a few yards difference, for example only flaggingfaster than 65.

As far as what to do in stretches of road which have them, I just engage cruise control and set it to the appropriate number.

Okell

facade said:

Okell said:

 <snip>
The first camera records your car when it goes past whatever part of the road the camera is aimed at.  The same with the second camera.

While the aiming point on the road must be some distance from the camera lens (I suppose it could reasonably be between perhaps 20m and 50m from the camera) it is a fixed point and won't move.  
<snip>

Since I have tried to research this, I am surprised that I can't find an answer as to what the fixed point is, where it is in relation to the camera and how they ensure that there is a measured distance between the points when they don't place a marker on the road, which would seem to be the logical thing to do, and then aim the camera at it.

It shouldn't be a secret, as knowing the answer won't give to a way to beat the cameras.

I think my curiosity has been piqued as much as yours, and I'd like to know the details as well!

I suspect the accurate measurement between the two fixed points presents few problems.  So far as I know, organisations like the Ordnance Survey use satellite photometry to produce and to revise pretty accurate maps.  If they can do that, I suspect Highways England (or whatever the relevant authority is) can measure pretty accurately the distance between two defined points - whether by satellite imagery or by physical surveying.

Over a sufficiently long base*, I would have thought an accuracy of +/- 0.3%** or better would easily be achievable

Insofar as to how the fixed points are referenced by the cameras, I suspect there actually are markings on the carriageway...

A quick search on PePiPoo (before it passes away) of "average speed" chucks up a lot of hits.  One of these is this one where, if you look at the photos from the entry camera and the exit camera, there appear(?) to be clear reference marks on the road.  (The shape of a letter "L" rotated 90 degrees clockwise).  NIP received A406 AVG Speed Camera - FightBack Forums (pepipoo.com)

I might be wrong, but it looks to me like the camera is targeted so that those two shapes are central in the image.  (I haven't looked at all the other search hits yet - most of them don't include photos unfortunately).

But from the point of view (no pun intended) of the question posed by the OP, it's all academic.  Questions like how far away can you be detected by average speed cameras and what is their range suggest that the OP - despite their protestations - doesn't understand the basic principle behind them.  How could a camera detect a vehicle passing a known fixed point at a specific time if the camera was a mile away from the fixed point?

[Edit:  That Pepipoo post also gives a "Declaration of Baseline Distances and Site Geometry", including camera height, horizontal distance of the camera from the datum (reference) point, and camera skew.  No idea what that is.  They also appear to claim to have baseline measurements to an accuracy of 1mm.  Or they may be taking the average of four such measurements - which seems reasonable]


*I think you mentioned a minimum distance of 200m in an earlier post.  That might very well be correct, but without knowing what levels of accuracy and precision are being used it seems a bit short to me

**  I think you also suggested in an earlier post that +/- 0.3% was the claimed level of accuracy

facade

Okell said:

facade said:

Okell said:

 <snip>
The first camera records your car when it goes past whatever part of the road the camera is aimed at.  The same with the second camera.

While the aiming point on the road must be some distance from the camera lens (I suppose it could reasonably be between perhaps 20m and 50m from the camera) it is a fixed point and won't move.  
<snip>

Since I have tried to research this, I am surprised that I can't find an answer as to what the fixed point is, where it is in relation to the camera and how they ensure that there is a measured distance between the points when they don't place a marker on the road, which would seem to be the logical thing to do, and then aim the camera at it.

It shouldn't be a secret, as knowing the answer won't give to a way to beat the cameras.

I think my curiosity has been piqued as much as yours, and I'd like to know the details as well!

I suspect the accurate measurement between the two fixed points presents few problems.  So far as I know, organisations like the Ordnance Survey use satellite photometry to produce and to revise pretty accurate maps.  If they can do that, I suspect Highways England (or whatever the relevant authority is) can measure pretty accurately the distance between two defined points - whether by satellite imagery or by physical surveying.

Over a sufficiently long base*, I would have thought an accuracy of +/- 0.3%** or better would easily be achievable

Insofar as to how the fixed points are referenced by the cameras, I suspect there actually are markings on the carriageway...

A quick search on PePiPoo (before it passes away) of "average speed" chucks up a lot of hits.  One of these is this one where, if you look at the photos from the entry camera and the exit camera, there appear(?) to be clear reference marks on the road.  (The shape of a letter "L" rotated 90 degrees clockwise).  NIP received A406 AVG Speed Camera - FightBack Forums (pepipoo.com)

I might be wrong, but it looks to me like the camera is targeted so that those two shapes are central in the image.  (I haven't looked at all the other search hits yet - most of them don't include photos unfortunately).

But from the point of view (no pun intended) of the question posed by the OP, it's all academic.  Questions like how far away can you be detected by average speed cameras and what is their range suggest that the OP - despite their protestations - doesn't understand the basic principle behind them.  How could a camera detect a vehicle passing a known fixed point at a specific time if the camera was a mile away from the fixed point?

[Edit:  That Pepipoo post also gives a "Declaration of Baseline Distances and Site Geometry", including camera height, horizontal distance of the camera from the datum (reference) point, and camera skew.  No idea what that is.  They also appear to claim to have baseline measurements to an accuracy of 1mm.  Or they may be taking the average of four such measurements - which seems reasonable]


*I think you mentioned a minimum distance of 200m in an earlier post.  That might very well be correct, but without knowing what levels of accuracy and precision are being used it seems a bit short to me

**  I think you also suggested in an earlier post that +/- 0.3% was the claimed level of accuracy

RichardD1970 posted an urban average camera set up near me, I used google and there are tiny calibration marks in the road around  30m or so from the cameras.

Those cameras are about 370m ( 1/4 mile ish) apart, there seems to be a chain of them, presumably multiple zones as naughty speeders could easily turn off or have to stop mid way through a section and not record an illegal average speed.

That data from pepipoo is interesting, the calibration point/datum is always 34m from the camera, but the cameras are different heights, which won't matter if you have a marked datum.

Skew is usually a reference to a tilt applied to get the x & y axes perpendicular.

So that answers our questions.

The camera software must have a narrow "hotspot" capture band across the image, and to set it up they just software adjust the centre of the band to co-inside with the marker (Like setting up alarm zones or privacy zones on a security camera), as soon as the numberplate enters the band it is recognised and the timer started or stopped.

I can't find where I got that 0.3% from (I should have bookmarked it)  but the accuracy would simply be the total width of the two capture bands divided by the distance between the cameras, if we assume they can time to .001 seconds the timing error would be negligible

The OP's question- how far from the camera does it detect me? is 

Around 30m

So provided there are no mobile cameras or Police about, you can pass the first camera at 180 if you like, as long as you ensure that you slow enough so that when you get to 30m before the second camera (if forward facing) it took you [3600 x (distance between cameras in miles)/(speed limit in mph)] seconds

Obviously 180 would break several laws, but the average speed cameras would not catch you- unless an officer viewed the live recording from the first camera and formed the opinion that you were clearly speeding......)

Nearlyold

Professional Surveyors GPS used in tandem in the baseline method are accurate to within 1mm per Km

HHarry

I think the point the OP makes is that if a forward facing camera is recording your speed half a mile away then you could theoretically speed up after that point and “safely” be doing more than the limit as you pass under the camera.

Equally if a rear facing camera records half a mile away then you need to maintain you speed as you pass under the camera and for a further half mile.

 If you know the detection points you can game the system by increasing / reducing your speed earlier or later rather than at the speed limit signs.  But the nett benefit is going to be negligible and the proper answer is of course “just follow the speed limit signs”.

Car_54

HHarry said:

I think the point the OP makes is that if a forward facing camera is recording your speed half a mile away then you could theoretically speed up after that point and “safely” be doing more than the limit as you pass under the camera.

It doesn't matter what speed you're doing as you pass that camera, nor indeed the next one. The cameras do NOT measure your speed.

As has been explained several times above, the speed is calculated by the time taken to pass between a pair of cameras.

Nearlyold

HHarry said:

I think the point the OP makes is that if a forward facing camera is recording your speed half a mile away then you could theoretically speed up after that point and “safely” be doing more than the limit as you pass under the camera.

Equally if a rear facing camera records half a mile away then you need to maintain you speed as you pass under the camera and for a further half mile.

 If you know the detection points you can game the system by increasing / reducing your speed earlier or later rather than at the speed limit signs.  But the nett benefit is going to be negligible and the proper answer is of course “just follow the speed limit signs”.

But if there's a third camera wouldn’t you then have to slow down to below the speed limit  to reduce your average speed otherwise the third camera will  pick you up as speeding. As you say seems an awful lot of bother for very little gain.

Car_54

Nearlyold said:

HHarry said:

I think the point the OP makes is that if a forward facing camera is recording your speed half a mile away then you could theoretically speed up after that point and “safely” be doing more than the limit as you pass under the camera.

Equally if a rear facing camera records half a mile away then you need to maintain you speed as you pass under the camera and for a further half mile.

 If you know the detection points you can game the system by increasing / reducing your speed earlier or later rather than at the speed limit signs.  But the nett benefit is going to be negligible and the proper answer is of course “just follow the speed limit signs”.

But if there's a third camera wouldn’t you then have to slow down to below the speed limit  to reduce your average speed otherwise the third camera will  pick you up as speeding. As you say seems an awful lot of bother for very little gain.

If there is a third (or fourth, or fifth) average speed camera, then the speeds are measured between the adjacent pairs, not over the whole section.