Safe Speed Forums

You'll find that posters on here generally try to avoid hyperbole. That's more than can be said for you and safety camera partnerships generally.

As stated above, it has been admitted that the LTI20:20 was not tested for incidence of slip error. The programme demonstrates what most of us already knew; that slip error can be readily reproduced. Any objective observer would not fail to conclude that it is now up to the authorities who assert that the device is wholly reliable in practice to back up that assertion - if they can - with further evidence. Failure to do so leads inevitably to the conclusion that the device is not wholly reliable. Then the only remaining question is - in a legal system which is founded on the presumption of innocence, how many unreliable and unsafe convictions/penalties are an "acceptable" cost weighed against the (claimed) benefits?

Its a tenuous defence really, especially as there is enough documentary evidence to show that (a) slip effect is not tested for in the Type Approval process (so they cannot claim it doesn't happen) and (b) the devices are in fact the same (thanks to the American manufacturers).

I think the US manufacturers would be very interested to hear that the UK Home Office is suggesting that their product is some way inferior to the TeleTraffic imported version. Maybe someone should send them a copy of the BBC program.

Does the following make sense?

1. As a matter of mathematics/physics which can be relatively easily calculated, where the target vehicle is at longer range (say 300m +), the degree of movement at the laser needed to cause a slip effect is tiny.

2. The iternal software will recognise a jerky or substantial panning movement (say if the beam jumps from one vehicle to another) and an error reading will be displayed.

3. If the panning movement is sufficiently small, the disparity between successive range results will not cause an error message. However, the speed error can still be substantial.

4. A conscientious operator will seek to minimise unintentional panning movement. However, particularly where the laser is handheld and possibly even if tripod mounted, the possibility of unintended movement cannot be eliminated.

5. Therefore, paradoxically, the more conscientious operators who seek to minimise incorrect readings could unintentionally be more likely to produce the slip effect and incorrect results seen in the programme.

We all know that it could happen.

Step up SCP's and show us the killer proof that it can't.

How do we all go about mounting a class action appeal?

Who do we write to?

JJ, Who do we write to to express our concern at this apparent miscarrage of justice?

if I remember my maths correctly from an age long, long ago then a shift of less than .5 of a degree of the laser will create a shift of over 2m at a range of 300m and 3m at 500m. So how exactly do you stop a laser moving less than .5 degree without a tripod on solid ground?

The maths is easy. It's simple trigonometry.

Let R be the range to the target.
Let O be the offset generated from the original aiming point when the device is moved through an angle A.

You get a right angled triangle, where R is the adjacent side, and O is the opposite side.

TAN A = O / R

O = R * TAN A

So for R = 300m, A = 1 degree, O = 5.2m.

The devices are designed to operate at distances of up to 1000m.

For R = 1000m, A = 1 degree, O = 17.45m (!!!!!!!)

And don't forget the surface being struck could be slanted. Consider a bonnet. There is a considerable difference in distance from front to back, compared with the vertical rise.

Lets say a car bonnet is 1m long. Lets say that the rise is 0.2m. Lets say the car is travelling at 30mph exactly. If during the 0.3 seconds that the reading is taken, the laser slips from the back to the front of the bonnet, then that is an extra 1m added to the real distance travelled. 30mph is 13.41 metres per second - which is 4.02m per 0.3 seconds. If we now add our slipped metre in, this gives 5.02m per 0.3 seconds. This in turn gives 16.74 metres per second - which is 37.5 mph.

So if this vertical slip downwards of 0.2m occurs, your speed gets exaggerated by 7.5mph!!!!!!!!

So what angle A is required to generate 0.2m of offset (O) at range R?

Well again,

TAN A = O / R

A = TAN^-1 (O/R)

For R = 300m

TAN A = 0.2 / 300

A = 0.038 degrees!

For R = 1000m

TAN A = 0.2 / 1000

A = 0.011 degrees!

So slip that could result in you being wrongly busted requires a movement of less than one tenth of one degree to happen.

With tripods or a handheld device the sligtest vibration - even from pressing the button itslef will be enough for this.

Scary isn't it!

Especially when the operator has to move it to use it.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

What also bothers me is that the cameras (even IF they are tripod mounted, and IF the operator is perfectly still and IF the operation of the trigger doesn't cause the gun to move) the cameras are located on overbridges.

These bridges are designed to move and flex as traffic moves over them. It is inherent in their construction, otherwise the rigidity of the concrete would cause them to crack and fail. Try standing on such a bridge as traffic goes past - it is an unnerving experience; I know of people who get similar feelings to being seasick.

Even IF all the above are adhered to, the flexibility in the bridges would cause the cameras to wobble all over the place and certainly more than the 0.1 degrees being quoted elsewhere.

I think the movement of a bridge is a minimal effect, but its a fair point.

We can show a wide range of factors that will induce the small movements necessary to create significant 'slippage' at distance.

Where do we go from here? this is more than a point to bang on to JJ about...this is a miscarrage of justice on a massive scale.

But empirical tests (including those in the programme) show that these devices do not give error messages or erroneous readings all the time so, despite the above, the internal software algorithms must be reasonably effective at trapping and neutralising the error potential described.

The crux of the issue, for me, is how often does the device fail to catch an error in real conditions (as opposed to controlled tests) and display a materially incorrect speed? If, as shown, it can fail to catch some errors, how many more are being missed. Is 99% reliability good enough? 99.9%? 99.99%?

[edit]
The truth of that, it appears to me, is that we do not know and, as Gareth said, there is no way of telling because the incidence of factors that could cause a spurious reading are essentially random.

The posts above show that minimal effects have huge results. My point is that the ground has been assumed to be stationary. It often isn't. If you've ever stood on a motorway bridge, you'll understand that bridge movement is significant.

Quite, the operator is trying to keep the laser on the front of the vehicle and in some locations the car is rounding a bend or travelling at a slight angle. The operators usually try and target vehicles as far away as possible (because they are less likely to have seen the van), so they are already panning the laser on a tiny object in the distance.

One thing that compounds the situation: If the operator is using the video screen to target vehicles (not the standard operating procedure) and the video camera isn't aligned with the laser then the operator could actually be unintentionally targetting the side of the vehicle.

Again, the various SCP's claim that they rigourously follow standard procedure but we have pretty much seen every rule in the book broken over the years.

Gareth

If you're going to take someone's licence away then nothing less than 100% reliability is good enough.

We're looking for 'beyond a reasonable doubt''. A one in ten million error would not normally make a 'reasonable' doubt.

My feeling is that the real world error rate probably lies in the range 1:25 to 1:250.

I hear a lot of indignant motorists saying: "I know I wasn't doing that speed." I believe a significant percentage of them.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

How many checks do they do per year? Probably more than that, no? That means at least 1 person a year will be incorrectly fined and potentially lose their licence and livelihood with absolutely no means of proving their innocence.

It is very good to see somebody with their thinking cap on! And this is a very good point. However maybe the answer is simple, and not what you might think intuitively.

While we've got some brains here, and the thickies are staying quiet, it's time to go up a gear.....

So far the assumption has been that these devices produce a spurious speed reading once in a while. Is this a good assumption?

First of all have a think about what the speed displayed actually means. It is not an instantaneous speed. In fact it's not even an accurate speed - there are no decimal points for a start.

The device uses a whole pile of analogue electronics to measure the round trip time of a beam of light. The target is moving and so there is no way the same point on the target will be hit for each pulse of the laser - even without slip. For each round trip of the laser a distance is calculated. From a series of such distances an **AVERAGE** speed is calculated, and displayed, without the decimal point.

The target could be accelerating or decelerating, and the angle being formed will be constantly changing. So it is clear that the displayed speed is only a "nominal" value. It would be impossible to show at what point during the 0.3s sample period that actual speed ocurred.

So it is clear that the accuracy isn't 100% even before we start worrying about jerks/slip etc.

I reckon that in actual fact the device NEVER reads the speed accurately but always reads an approximation. However the DEGREE of accuracy varies, depending on a number of factors, including jerking and slipping etc.

This means that often the device is "near enough" and gross, immediately obvious errors are RARE. Note that "RARE" does not equal "NEVER HAPPENS".

One of the factors will obviously be the accuracy of the targeting, and clearly misaligned devices are a disaster waiting to happen. But how exactly accurate can you target a device? Is there such a thing as a perfectly aligned device? I don't think there is, because there is enough parallax between the red dot aiming mechanism on the LTI 20.20 to mean that if your eye position shifts relative to the red dot then the aiming point at distance varies by a lot more.

There is some more evidence of this. I have seen static objects get speeds of 1 or 2mph. But the key one is that I have seen readings of -0mph as well as 0mph. For this to happen, the device must be seeing some sort of difference. Imagine the decimal point put back in. I would be prepared to bet it would NEVER measure exactly 0.000mph.

This is why for static targets sometimes the LTI 20.20 displays 0mph, and sometimes it displays -0mph.

Here's how the "expert" from TeleTraffic describes it, describing an instance of this, picked up by a Judge:



Perhaps the software rolls a dice, or tosses a coin.

And if the camera is in a van at the side of the road, how much effect is there from buffeting from passing lorries, etc?
More than a couple of degrees, I'll wager.

Cheers
Peter

_________________
Only when ideology, prejudice and dogma are set aside does the truth emerge - Kepler

I'd certainly like it to be perfect. I'd like it if no innocent people were ever convicted of anything. But in the real world it's going to happen - innocent people will be found guilty. We have to minimise the chances - we can't eliminate them because there's no such thing as perfection.

But in the case under consideration, I believe that the probability of error far far exceeds such levels. In fact I believe that there's gross injustice going on.

_________________
Paul Smith
Our scrap speed cameras petition got over 28,000 sigs
The Safe Speed campaign demands a return to intelligent road safety

true, but I think we're looking at it from different directions as it were.
What I'm really saying is that I want the tolerance on the machine to be something like +- 1% (or whatever) in all cases rather than 'the speed is accurate but every now and then it will read 100mph instead of 70'.

Especially considering the nature of the alleged offence.

Cheers
Peter

_________________
Only when ideology, prejudice and dogma are set aside does the truth emerge - Kepler