Race start monitoring system

All that has to do with the power train, gearbox, clutch, fuels and lubricants, etc. Generally the mechanical side of Formula One.
AJI
AJI
27
Joined: 22 Dec 2015, 09:08

Re: Race start monitoring system

Post

deltaecho5 wrote:
15 Jul 2017, 14:46


On-Board camera operation is not my speciality, I am familiar only with cameras used in the operation of timekeeping, and I am simply not at liberty (or willing due to provocation) to divulge the complete operation of FIA procedure, equipment and computations used to determine the officiating of a race start.

A reasonable person could conclude using data extrapolated from video systems, on-board telemetry and timing systems to provide an aggregate measurement if a competitor had a false start. This is the current process.

I have not indicated there are GPS systems on an F1 car used for the purposes of start box movement verification.

Again, to emphasize my reply to the assumptions that had been made: there are no loops in the start boxes of F1 tracks. If the information I have provided is of no use to the thread, my apologies...
I have suggested three completely independent data sources for the 'binary' determination of whether a jump-start has occurred but you cannot confirm the resolution of any of them? It's not like that information would be sensitive... I'm not asking you to divulge the secrets of the FIA, I'm simply asking if you can confirm that the 'broad view' camera (that you have suggested is the primary data source for determining jump-starts) runs at 1000fps or more?

I'm happy to add car telemetry to the list of acquired data for verification. What is the specific data collected from telemetry that would assist in the decision? IMO it can really only be wheel speed and GPS data. Perhaps brake pedal pressure could be taken into account, but it can't determine whether a car has moved or not.

At no stage have I suggested that induction loops at each car's start position have been used for the determination of whether a jump start has occurred, because quite frankly I think there is enough data from the three sources I have listed above to achieve this, however I applaud RZS10 for their work on trying to fill the vacuum that has been left by the FIA and FOM in this regard.

User avatar
deltaecho5
33
Joined: 26 Jun 2017, 03:54
Location: Usually at the Track

Re: Race start monitoring system

Post

hardingfv32 wrote:
15 Jul 2017, 15:00
With your system does the position of the are car (distance from the transponder to the center of the loop) create a signal that varies depending on the distances? Could it not be stated that these signal/distance relationships form a curve?
The transponder hits relative to the loop are signal strength over time, in a scatter graph to form a curve, the Tx hits are on the left of the graph as strength, time is represented on the bottom. In the administration of the decoder system, this data is represented as such.

If a Tx hit is high, over time, the representation is more of a straight line. This is typically the case of a Tx that is sitting on or near a loop. Movement up in strength, or peak signal, is represented upwards as a curve and then downwards (the car approaching the loop, passing over the loop and traveling away from the loop).

For the sake of a fictitious system to be used:
Various methods using a Tx system could be employed (their installation would have to be permanent in nature, as all other track loops) to determine the Tx position to the installed loop. Since the sensitivity of the Tx to loop reception is ample, it would make sense the loop would have to be a sufficient distance away from the Tx to disregard hits (not receive them), and only sense them when the Tx approached a predetermined limit or distance to the loop. Otherwise the system would be a max peak constantly and discerning the difference would be obtained by changing the squelch settings, and turning down the hits registration. The system isn't that reliable currently, since it's designed to accept anything from the Tx rather than discern the signal strength over time.

The signal strength is only used to determine the peak hit, and ensure the Tx is operational. If only one hit is received, regardless of it's strength, it's sufficient to show the car passed the loop. This is what is known as scoring, and a car can only be scored if it passes a scoring loop. There is only one scoring loop on the track, and that is known as the Start Finish loop. A competitor cannot be deemed to have started the race until they have passed the Start Finish loop. That is what the Tx system is designed to recognize, not their relative position to the S/F or their start box, or their relative position to other competitors positions or their competitors positions in their start boxes.

Using Tx's in this manner would be like using a photocell from a binary operation of determining if a car passed the S/F to trying to get data on how much light the car gave off as it passed by.

There are other means to determine a false start, and those are employed by the FIA from sources that allow aggregate data to make a binary decision. Telemetry is obtained by the FIA, and in this method, (not GPS) the data is used to determine absolute stop/go, both on the start line and also in the pit box.

For the sake of the discussion, I cannot address the specifics of why a specific reaction time was mentioned.
⏱ Timekeeper by day, F1 Driver by night (whilst sleeping of course...)

User avatar
deltaecho5
33
Joined: 26 Jun 2017, 03:54
Location: Usually at the Track

Re: Race start monitoring system

Post

hardingfv32 wrote:
15 Jul 2017, 15:10
Thanks for your information. I was having difficulty developing a jump start system that made sense with technical reality. So what we have is a system using human judgement from multiple data source. The statement of high accuracy numerical reaction times etc. are implying a sense of accuracy that does not exist.
Not to beat a dead horse, but yes, exactly! Please understand my position is not to contradict, or by provision of information allow the conclusion of contradiction of the edicts, decisions or promulgation of information from the FIA. In this case I can only address what it isn't.

Given enough money and time there is always a way to do something, but whether that something is necessary isn't always clear. F1 is a circus of somethings because of the level of technical proficiency of the competitors, the money spent and the money at stake.

In timekeeping, there is always a push to utilize currently capitalized equipment to yield increasingly broader and more accurate results.

When we have exhausted the technical limits of the equipment, we either buy more equipment and develop a new protocol or we resolve to use other methods that yield results acceptable to the competitors.

BTW, from a timekeeping prospective: the winner is the competitor that covers the greatest distance in the shortest time. The only time this would be an issue for scoring is if Bottas finished less than 201ms ahead of the nearest competitor at the conclusion of the race.
Last edited by deltaecho5 on 15 Jul 2017, 16:12, edited 1 time in total.
⏱ Timekeeper by day, F1 Driver by night (whilst sleeping of course...)

3jawchuck
3jawchuck
37
Joined: 03 Feb 2015, 08:57

Re: Race start monitoring system

Post

AJI wrote:
15 Jul 2017, 14:20
.
...Where does the 201ms figure come from? Is the 'broad view' camera that references the lights-out time stamp running at 1000fps? If it is I find it hard to believe it can capture a high resolution view of the first 4 rows let alone the entire field...
...
I would like to know this too. What frame rate do the onboard cameras use? I guess this could be combined with telemetry to give a fuller picture. Collecting this data, making sure it's all running in the right time frame and analysing it would probably explain the time it took to get a verdict on the start.
Last edited by 3jawchuck on 16 Jul 2017, 01:56, edited 1 time in total.

User avatar
RZS10
359
Joined: 07 Dec 2013, 01:23

Re: Race start monitoring system

Post

hardingfv32 wrote:
15 Jul 2017, 15:00
With your system does the position of the are car (distance from the transponder to the center of the loop) create a signal that varies depending on the distances? Could it not be stated that these signal/distance relationships form a curve?
Assuming such a system would be in use ...yea ... you can use 'create' or 'form' interchangeably ... the car would drive towards the sensor, at some point it would start picking up the signal from the transmitter, the signal strength would go from zero to the horizontal line you can see at the beginning of every sketch, to that you'd add that tolerance ...

But i think it's time we stop discussing a system that might not even be in use :-k

AJI
AJI
27
Joined: 22 Dec 2015, 09:08

Re: Race start monitoring system

Post

RZS10 wrote:
15 Jul 2017, 20:33

But i think it's time we stop discussing a system that might not even be in use :-k
As we are getting no detail on what system is used it seems we have to go back to speculation...

Assuming what deltaecho5 has said is true (and that's a pretty big leap of faith for me) we can assume the following:

- The 'broad view' camera must run at a minimum of 1000fps to be able to deliver a 1ms resolution (but we can't get confirmation on that)
- The 'broad view' camera must have AMAZING optics to be able to see the whole field (but we can't get confirmation on that)
- The time stamp system must be synchronised to the 'broad view' camera (phew, there's one we can probably lock down!)
- Car telemetry is used, but we don't know what telemetry or what resolution (because we can't get confirmation on that)
- On board car cameras may be used (but we can't get confirmation on frame rate)

What we can probably discard is:
- Your theory of induction loops in the grid boxes (although I thank you for your excellent work)
- GPS data (as even though we can't get an answer on sampling rate I have to assume it's way to low to be useful)

So, I'd call that a big-fat-nothing.
I'm calling Ross Brawn

AJI
AJI
27
Joined: 22 Dec 2015, 09:08

Re: Race start monitoring system

Post

deltaecho5 wrote:
15 Jul 2017, 15:47

BTW, from a timekeeping prospective: the winner is the competitor that covers the greatest distance in the shortest time. The only time this would be an issue for scoring is if Bottas finished less than 201ms ahead of the nearest competitor at the conclusion of the race.
The implication of the above statement is that BOT jumped the start by 201ms, is it not? Can we at least get a confirmation on that? Pretty please with sugar on top?

hardingfv32
hardingfv32
35
Joined: 03 Apr 2011, 19:42

Re: Race start monitoring system

Post

RZS10 wrote:
15 Jul 2017, 20:33
Assuming such a system would be in use ...yea ... you can use 'create' or 'form' interchangeably ... the car would drive towards the sensor, at some point it would start picking up the signal from the transmitter, the signal strength would go from zero to the horizontal line you can see at the beginning of every sketch, to that you'd add that tolerance ...
Our proposed system is much more interesting than a bunch of stewards watching videos.

So back to your statement: I would think the as the car drives 'towards' the sensor that the signal strength would go from zero to peak fallowing a tradition parabolic signal strength/distance curve. Can you explain why you feel it would go horizontal? There might be something for me to learn.

Brian

User avatar
deltaecho5
33
Joined: 26 Jun 2017, 03:54
Location: Usually at the Track

Re: Race start monitoring system

Post

hardingfv32 wrote:
16 Jul 2017, 01:09
So back to your statement: I would think the as the car drives 'towards' the sensor that the signal strength would go from zero to peak fallowing a tradition parabolic signal strength/distance curve. Can you explain why you feel it would go horizontal? There might be something for me to learn.
I'll take another shot at getting some info out about this question (an excellent question BTW, please excuse my answers being to the point not because of sarcasm or temper, but due to the amount of time it takes to type this in between breaks). BTW, I stand by Salo and Whiting as to the tolerances being adhered to regarding Bottas.

Refresher-
-No loops in the boxes (see below)
-Onboard telemetry (think g/motion sensor), the camera's are useful, as shown to the public.
-There is more than one high-res/high speed broadview camera (why do you think there would be one to see the whole field??)
-The start system uses a timestamp which (careful...the word GPS shows up here) uses GPS.

Why GPS? The GPS system utilizes significant complex mathematical formulas to calculate synchronous orbits...all to do one thing: Give correct positioning information. The by product of this is accurate timing. TIMING, not location. Virtually of all of race operations equipment utilizes this timestamp. When GPS is referenced when relative to timing, it's the GPS timestamp, not what people think of using a Garmin...the location data available for non-military use isn't accurate enough for motorsports.

(If you haven't read my post on how timing works, you may find it interesting:

viewtopic.php?f=1&t=9114&p=704403#p703625

OK, so let's look at the timing loop/Tx Signal information. (Buckle up, this is going to be a long and boring post to address how the system works, and I will let you draw your own conclusions as to what it's usefulness would be for various applications other than passings logging.)

Imagine a fixed loop in the track (pick your poison...pit lane, data point, sector loop, S/F, or even for the sake of this discussion, the start box (grrrrrr)).

The loop is a receiver of a digital signal transmitted by the onboard Tx. It is at a fixed point on the ground, embedded in a cut and caulked channel in the track surface able to sustain high speed traffic. See the images previously posted of the S/F line.

Let's look at a normal passing on track first-
If a competitor is traveling over the loop in race direction, the signal of the Tx will start out as a weak hit (one data point on a graph at the lowest signal, arriving at a specific timestamp...I have scoured my Mac for a picture of this, but cannot find any I've my fingertips on...maybe at some future point I can post the exact system picture to ease your minds.

As the Tx moves towards the loop, the signal strength (SS) increases and another data point, another hit is registered (this actually takes place in milliseconds, so it's actually "hits"). As the Tx closes on the loop, the graph would continue upward as discussed. Until the Tx is in such a position that it reaches its maximum SS in relationship to the loop. The timestamp of that hit is compared with all other hits from that specific TX and it is logged as the actual passing. As the Tx passes over and travels beyond the loop in race direction, the signal begins to degrade, and each hit (timestamp) begins being marked on the downward slope of the graph in relation to time.

Here's a representation:
Image

Now let's look at what happens when a car is parked/stopped/resting over the same loop (this happens in the pit lane and on track as a matter of racing).

The car approaches the loop, the SS slowly increases. As the car sits, the loop receives maximum hits (the TX transmits it's digital signal constantly, so the hits are many, and the SS of each can vary due to normal issues with any receiver. Again, the system is designed to "listen" for digital signal on the Fq of the TX's and by default, hears anything on that Fq and timestamps it relative to the Tx's specific serial number.

So while the Tx is sending signal, the system has a setting to show Max signal and low signal (this is set for the system, the loop, not by Tx. Squelch is also set by the loop and decoder, not by Tx. But let's just say that you could distinguish the peak signal. A car coming to rest near a loop would have a SS if it maintains it's location. As it progressed to the loop, the signal strength would increase if it moved forward. As it traveled forward, the SS would be represented in a series of lines in upward plateaus. If the car were to back away form the loop, the SS would decline, and the timestamps on the graph would look as if the car had travel over the loop and beyond it, as above. There would be no way to determine direction in this way, only SS over time. Cars rarely pass over a loop on track and back up once they hit the center of the loop (the max SS) at speed...this would be pretty rough on the gear box...

A resting car in a pit box scenario or sitting on a loop looks like this:

Image

There is no way to determine the cars direction relative to the loop, although this method could be used to determine it's proximity to the loop.

The problem is each cars Tx has a different and varying SS, and the loops ability to pick up the signal also generates anomalies and outliers, and this is where the timestamps get irrelevant. So a car standing still could and often will have an increased SS which generates a hit and time stamp, which would show as a movement even though the car was standing still.

IF utilizing the Tx relative to a cut loop in the track to sense a start box false start, it would ultimately only be useful to show the car moving over the loop...inch by inch movement would be difficult to prove. Given this, the video evidence and telemetry ("sensor system") connected to the track timing network (track sensors) is a more reliable way to determine if a competitor has broken the rules.

In any event, they are allowed to react, or in some cases pre-judge the lights out and launch. They suffer a penalty when this movement is deemed to have occurred outside the acceptable norm of movement from the car for other reasons.
⏱ Timekeeper by day, F1 Driver by night (whilst sleeping of course...)

AJI
AJI
27
Joined: 22 Dec 2015, 09:08

Re: Race start monitoring system

Post

deltaecho5 wrote:
16 Jul 2017, 04:14

Refresher-
-No loops in the boxes (see below)
What is that in the grid box?

Image

User avatar
deltaecho5
33
Joined: 26 Jun 2017, 03:54
Location: Usually at the Track

Re: Race start monitoring system

Post

AJI wrote:
16 Jul 2017, 04:18
deltaecho5 wrote:
16 Jul 2017, 04:14

Refresher-
-No loops in the boxes (see below)
What is that in the grid box?

http://www.grandprix247.com/wp-content/ ... 60x240.jpg
The backup S/F loop.
⏱ Timekeeper by day, F1 Driver by night (whilst sleeping of course...)

AJI
AJI
27
Joined: 22 Dec 2015, 09:08

Re: Race start monitoring system

Post

deltaecho5 wrote:
16 Jul 2017, 04:32
AJI wrote:
16 Jul 2017, 04:18
deltaecho5 wrote:
16 Jul 2017, 04:14

Refresher-
-No loops in the boxes (see below)
What is that in the grid box?

http://www.grandprix247.com/wp-content/ ... 60x240.jpg
The backup S/F loop.
Phew! I thought this was about to start all over again... Thank you for answering so swiftly.

User avatar
deltaecho5
33
Joined: 26 Jun 2017, 03:54
Location: Usually at the Track

Re: Race start monitoring system

Post

LOL...I was hoping not.

Here's a snap of the typical trackside box for S/F...you'll see the different BNC conx for the loops...in this case it's the DIS track (NASCAR/Rolex/24hr Sportcars). There are typically 4 at S/F, two S/F, two S/F Pits. We connect decoders to all 4 but only compile data from the mains unless something happens. The distance and location of the back-ups varies from track to track, but this is accounted for in the software.

At FIA tracks, the loops still use the same system. At FE and street circuits, we use a portable fiber ring. In this snap, these are just the S/F loops under the start stand...image if you had 20 loops at grid? Yikes. I realize it's tough to believe, but in my job I come across everything at the tracks around the world, for each sanctioning body, each type of race...there's little there we don't touch or have to repair...I'd remember 20 extra loops :)

Image


Image
⏱ Timekeeper by day, F1 Driver by night (whilst sleeping of course...)

AJI
AJI
27
Joined: 22 Dec 2015, 09:08

Re: Race start monitoring system

Post

deltaecho5 wrote:
16 Jul 2017, 05:13
LOL...I was hoping not.

Here's a snap of the typical trackside box for S/F...you'll see the different BNC conx for the loops...in this case it's the DIS track (NASCAR/Rolex/24hr Sportcars). There are typically 4 at S/F, two S/F, two S/F Pits. We connect decoders to all 4 but only compile data from the mains unless something happens. The distance and location of the back-ups varies from track to track, but this is accounted for in the software.

At FIA tracks, the loops still use the same system. At FE and street circuits, we use a portable fiber ring. In this snap, these are just the S/F loops under the start stand...image if you had 20 loops at grid? Yikes. I realize it's tough to believe, but in my job I come across everything at the tracks around the world, for each sanctioning body, each type of race...there's little there we don't touch or have to repair...I'd remember 20 extra loops :)
Thanks. I read the other stuff you posted on the timing thread. Very interesting and all very clear.

Now, I'm sorry to harp on about this, but I'm still unclear about BOT's "anticipated" start, so can you please help?
I accept that his start was within the (un-specified but approx. -200ms) limits which implies that he started before the lights, but the 201ms number is still reported as an RT. The term Reaction Time implies that he reacted to the lights, specifically, 201ms after the lights, but we know from video footage that he did indeed move before the lights went out. Could this simply have been reported incorrectly? i.e. That he launched 201ms before the lights and someone accidentally appended the term RT to the number? I know you can't answer the question definitively as you weren't there, but the suggestion that 201ms is an RT for someone who was moving before the lights went out is directly contradictory.

Cold Fussion
Cold Fussion
93
Joined: 19 Dec 2010, 04:51

Re: Race start monitoring system

Post

AJI wrote:
15 Jul 2017, 14:20
What is the frame rate of the onboard cameras? Assuming the on-board cameras run at 100fps we can see whether a jumped start was made to a 10ms resolution. If this is the case then why not use that system?
It's unlikely the onboard cameras run at 100 fps, it's going to be either 50 or 25.