GPS data for track creation

Reading up on the new Indy game from Codemasters at BHMS, I see they write that GPS data is "too inaccurate" to use as track data. They quote up to 30cm off. Well, for your out-of-the-store handheld GPS unit this is true. I've worked with a friend who runs a company dealing with on/offshore GPS products, and by using his setup/systems and a radio link base station, we get down to a discrepancy of 1-2 cm. We can easily measure roads with incredible detail (20Mhz) by mounting 4 tiny antennas on a car or a trolley. In other words, if we pulled a trolley after a car around a racetrack a given number of laps, and shook the data about to get rid of the overlaps, we would have a seriously accurate bucket of information. I have mentioned this to one of the upcoming sim (maybe upcoming?er..) companies with no reply. Initially we thought about making a plugin for say, 3dsmax, so that we could dump all the data straight over.These GPS systems are serious - and very expensive (in the �30-50000 bracket) and are used among other things to guide oil tankers etc so they dont whack the quays to pieces. Accuracy is paramount! Any thoughts? /Matt

Yeah, I work in an engineering department and we've used devices that are within 25mm (1 inch) accuracy and cost about US$15000. I think 30cm accuracy would be more than good enough for decent track creation.

Right there with you man. You cant blame the technology to be too inaccurate. There was conversation on this subject year ago. You might wanna check this thread: http://groups.google.com/groups?hl=fi&lr=&ie=ISO-8859-1&q=%22GPS+data+%26+accurate+circuit+modelling%22&btnG=Google-haku&meta= Cheers, Leo "Matt Knutsen" <matt|||SPAMGUARD||@knutspeed.com> wrote in message news:<b9sogu$o37$[email hidden]>...

Quoted text (24 lines)

> Reading up on the new Indy game from Codemasters at BHMS, I see they write > that GPS data is "too inaccurate" to use as track data. They quote up to > 30cm off. > > Well, for your out-of-the-store handheld GPS unit this is true. I've worked > with a friend who runs a company dealing with on/offshore GPS products, and > by using his setup/systems and a radio link base station, we get down to a > discrepancy of 1-2 cm. > > We can easily measure roads with incredible detail (20Mhz) by mounting 4 > tiny antennas on a car or a trolley. In other words, if we pulled a trolley > after a car around a racetrack a given number of laps, and shook the data > about to get rid of the overlaps, we would have a seriously accurate bucket > of information. > > I have mentioned this to one of the upcoming sim (maybe upcoming?er..) > companies with no reply. Initially we thought about making a plugin for say, > 3dsmax, so that we could dump all the data straight over.These GPS systems > are serious - and very expensive (in the �30-50000 bracket) and are used > among other things to guide oil tankers etc so they dont whack the quays to > pieces. Accuracy is paramount! > > Any thoughts? > /Matt

"Matt Knutsen" <matt|||SPAMGUARD||@knutspeed.com> wrote in message news:b9sogu$o37$[email hidden]... > Reading up on the new Indy game from Codemasters at BHMS, I see they write > that GPS data is "too inaccurate" to use as track data. They quote up to > 30cm off. > > Well, for your out-of-the-store handheld GPS unit this is true. I'veworked > with a friend who runs a company dealing with on/offshore GPS products,and > by using his setup/systems and a radio link base station, we get down to a > discrepancy of 1-2 cm. > > We can easily measure roads with incredible detail (20Mhz) by mounting 4 > tiny antennas on a car or a trolley. In other words, if we pulled atrolley > after a car around a racetrack a given number of laps, and shook the data > about to get rid of the overlaps, we would have a seriously accuratebucket > of information. > > I have mentioned this to one of the upcoming sim (maybe upcoming?er..) > companies with no reply. Initially we thought about making a plugin forsay, > 3dsmax, so that we could dump all the data straight over.These GPS systems > are serious - and very expensive (in the �30-50000 bracket) and are used > among other things to guide oil tankers etc so they dont whack the quaysto > pieces. Accuracy is paramount! > > Any thoughts? > /Matt > Sounds like a great idea (especially the max plugin), perhaps they are put off by the hefty price tag, although you would imagine that hiring it would be an option. Also gaining access to some tracks may be a little tricky. If I was them I'd jump at the chance to get some accurate tracks into max with little modelling effort. Hopefully it would be a case of adding the textures and off ya go, although I guess thats the most important part :). Most developers probably have the attitude that most people wont notice and whats it matter if things are a foot out here and there to the average punter on the street. Perhaps you should start a company and sell the data to multiple developers - i'll of course provide a test-bed in sirocco :))) And if you need me to do some measuring in Brazil / Oz / Italy / Spain...well anywhere...i'm ur man :)) All the best, Ash http://www.siroccoracing.com

"Ashley McConnell" <[email hidden]> skrev i melding news:b9tdp1$4u7$[email hidden]... > Sounds like a great idea (especially the max plugin), perhaps they are put > off by the hefty price tag, although you would imagine that hiring itwould > be an option. That was our idea. We could travel with a briefcase of GPS equipment and do some serious data logging using a vehicle from AVIS or any other rental outlet (that wouldn't mind the car returned with worn out tires, brake pads and shocks of course...heheh) Also gaining access to some tracks may be a little tricky. > If I was them I'd jump at the chance to get some accurate tracks into max > with little modelling effort. Hopefully it would be a case of adding the > textures and off ya go, although I guess thats the most important part:). Well, I believe it could be done. There are so many brilliant codies hanging around here - and I think most of them would jump at such a project. > Most developers probably have the attitude that most people wont noticeand > whats it matter if things are a foot out here and there to the average > punter on the street. Yeah, but then again - some would mind. And who could say no to 1) Be able to sell your track data to any interested software developer (track owner) 2) Be able to sell your data to race teams for testing/simulations 3) Drop the entire measuring process of building a track for a sim If you could buy the data for track creation (say that covers 70% of the development work) it's just a matter of seeing what would be cheapest. And it would also cut dev time dramatically. > Perhaps you should start a company and sell the data to multiple > developers - i'll of course provide a test-bed in sirocco :))) Hehe, if we do I'd love to. Scirocco rocks! We might do a test at my local track (Rudskogen) later this summer... > And if you need me to do some measuring in Brazil / Oz / Italy / > Spain...well anywhere...i'm ur man :)) It'll be us, our antennas and you of course! Sounds nice with some sun ;) > All the best, > Ash > http://www.siroccoracing.com Best /Matt www.knutspeed.com

"Leo Merikallio" <[email hidden]> skrev i melding news:[email hidden]... > Right there with you man. You cant blame the technology to be too > inaccurate. There was conversation on this subject year ago. You might > wanna check this thread: > >http://groups.google.com/groups?hl=fi&lr=&ie=ISO-8859-1&q=%22GPS+data+%26+ac curate+circuit+modelling%22&btnG=Google-haku&meta= > > Cheers, > Leo Thanks! -- Matt Knutsen www.knutspeed.com

http://groups.google.com/groups?hl=fi&lr=&ie=ISO-8859-1&q=%22GPS+data... curate+circuit+modelling%22&btnG=Google-haku&meta= Thanks! -- Matt Knutsen www.knutspeed.com

>"Damien Smith" <[email hidden]> wrote > Yeah, I work in an engineering department and we've used devices that are > within 25mm (1 inch) accuracy and cost about US$15000. I think 30cm > accuracy would be more than good enough for decent track creation. 30cm? Imho, having a corner with a radius off 30cm, can change the corner completely. -- -- Fran�ois M�nard <ymenard> -- http://ymenard.cjb.net/ -- This announcement is brought to you by the Shimago-Dominguez Corporation - helping America into the New World...

Surely you mean 30 mm? 30 cm error in the crown of a road or in the banking of a curve would make it drive dramatically differently. "Damien Smith" <[email hidden]> wrote in message news:b9srp0$n0htj$[email hidden]... > Yeah, I work in an engineering department and we've used devices that are > within 25mm (1 inch) accuracy and cost about US$15000. I think 30cm > accuracy would be more than good enough for decent track creation. > >

On Wed, 14 May 2003, Matt Knutsen wrote: > Yeah, but then again - some would mind. And who could say no to > 1) Be able to sell your track data to any interested software developer > (track owner) > 2) Be able to sell your data to race teams for testing/simulations > 3) Drop the entire measuring process of building a track for a sim Sorry to be the one to tell you, there are already one or more sources for high quality track data. Latest quote I heard was about US$1000/track, plus something up front for the software to interpret the data. Quite reasonable for a top level team or manufacturer. The accuracy requirement for the vertical dimension is very tight, a 5mm bump in the track is a big deal for an F1 car. 1mm vertical would be desirable.

> Surely you mean 30 mm? 30 cm error in the crown of a road or in thebanking > of a curve would make it drive dramatically differently. Sure, but are you telling me that there's a single game out there that has every corner within 30cm accuracy - please don't say GP4.

"Doug Milliken" <[email hidden]> skrev i melding

Quoted text (10 lines)

> > Yeah, but then again - some would mind. And who could say no to > > 1) Be able to sell your track data to any interested software developer > > (track owner) > > 2) Be able to sell your data to race teams for testing/simulations > > 3) Drop the entire measuring process of building a track for a sim > > Sorry to be the one to tell you, there are already one or more sources for > high quality track data. Latest quote I heard was about US$1000/track, > plus something up front for the software to interpret the data. Quite > reasonable for a top level team or manufacturer.

Very cheap imo - but that's solely guided towards telemetry data, not for a sim/game? > The accuracy requirement for the vertical dimension is very tight, a 5mm > bump in the track is a big deal for an F1 car. 1mm vertical would be > desirable. I wasn't thinking I had invented the wheel here:) But - if that's how close they need it surely the data must be renewed at an extreme interval? The ground moves about from a lot of factors, temperature, sifting and shifting and so on. I don't believe it would be possible to get it that close "out of the box". It is interesting to see where this is leading if one thinks of it as purely a way of simulating for "real cars". Would the FIA ban it? What about all the other factors? Would the tracks end up as a "Matrix" test bed with sensors measuring everything to give the simulators enough food? When would it actually be cheaper to take your F1 team there to do some real testing? Come to think of simulations, we built a C2 car in the late 80s and took it to a company in the UK called GeoScan who had it on the floor for a week. They measured all suspension points, weighed every nut and so on, ran the data through their software and gave us a book saying "move the rear lower wishbone mount up 2mm, front 1,5 mm" etc etc. After this the car would supposedly be 90% sorted. It wasn't bad at all after we did the changes - we did have to increase front track as well to counter some problematic lowspeed understeer but overall I am not sure just how much help that simulation was... /Matt

Surely the way GPS works is to move the entire data set by a variable amount over a period of time, right? So if the ends match up, the accuracy between successive points should be to all extents and purposes the maximum achievable anyway. Les Neilson "Matt Knutsen" <matt|||SPAMGUARD||@knutspeed.com> wrote in message news:b9sogu$o37$[email hidden]... > Reading up on the new Indy game from Codemasters at BHMS, I see they write > that GPS data is "too inaccurate" to use as track data. They quote up to > 30cm off. > > Well, for your out-of-the-store handheld GPS unit this is true. I'veworked > with a friend who runs a company dealing with on/offshore GPS products,and > by using his setup/systems and a radio link base station, we get down to a > discrepancy of 1-2 cm. > > We can easily measure roads with incredible detail (20Mhz) by mounting 4 > tiny antennas on a car or a trolley. In other words, if we pulled atrolley > after a car around a racetrack a given number of laps, and shook the data > about to get rid of the overlaps, we would have a seriously accuratebucket > of information. > > I have mentioned this to one of the upcoming sim (maybe upcoming?er..) > companies with no reply. Initially we thought about making a plugin forsay, > 3dsmax, so that we could dump all the data straight over.These GPS systems > are serious - and very expensive (in the �30-50000 bracket) and are used > among other things to guide oil tankers etc so they dont whack the quaysto > pieces. Accuracy is paramount! > > Any thoughts? > /Matt > >

On Thu, 15 May 2003, Matt Knutsen wrote:

Quoted text (25 lines)

> > "Doug Milliken" <[email hidden]> skrev i melding > > > Yeah, but then again - some would mind. And who could say no to > > > 1) Be able to sell your track data to any interested software developer > > > (track owner) > > > 2) Be able to sell your data to race teams for testing/simulations > > > 3) Drop the entire measuring process of building a track for a sim > > > > Sorry to be the one to tell you, there are already one or more sources for > > high quality track data. Latest quote I heard was about US$1000/track, > > plus something up front for the software to interpret the data. Quite > > reasonable for a top level team or manufacturer. > > Very cheap imo - but that's solely guided towards telemetry data, not for a > sim/game? > > > The accuracy requirement for the vertical dimension is very tight, a 5mm > > bump in the track is a big deal for an F1 car. 1mm vertical would be > > desirable. > > I wasn't thinking I had invented the wheel here:) But - if that's how close > they need it surely the data must be renewed at an extreme interval? The > ground moves about from a lot of factors, temperature, sifting and shifting > and so on. I don't believe it would be possible to get it that close "out of > the box".

Sorry, poorly stated -- it's "local accuracy" that needs to be very good, so that bumps are represented accurately. A "drift" in altitude from one end of the track to another should not be a problem. > It is interesting to see where this is leading if one thinks of it as purely > a way of simulating for "real cars". Would the FIA ban it? What about all > the other factors? Would the tracks end up as a "Matrix" test bed with > sensors measuring everything to give the simulators enough food? When would > it actually be cheaper to take your F1 team there to do some real testing? An early attempt would be Chaparral's Rattlesnake Raceway. These days Fiorano(sp?-Ferrari) must be one of the most heavily instrumented tracks?

"Doug Milliken" <[email hidden]> skrev i melding > > I wasn't thinking I had invented the wheel here:) But - if that's how close > > they need it surely the data must be renewed at an extreme interval? The > > ground moves about from a lot of factors, temperature, sifting andshifting > > and so on. I don't believe it would be possible to get it that close"out of > > the box". > > Sorry, poorly stated -- it's "local accuracy" that needs to be very good, > so that bumps are represented accurately. A "drift" in altitude from one > end of the track to another should not be a problem. ok, I was more thinking about the track changing over time. Doing a +/- 1mm check one day will probably not match the track the next, not to mention the next year. Even temperature should make the track surface heights fluctuate more than 5mm? > > It is interesting to see where this is leading if one thinks of it aspurely > > a way of simulating for "real cars". Would the FIA ban it? What aboutall > > the other factors? Would the tracks end up as a "Matrix" test bed with > > sensors measuring everything to give the simulators enough food? Whenwould > > it actually be cheaper to take your F1 team there to do some realtesting? > > An early attempt would be Chaparral's Rattlesnake Raceway. These days > Fiorano(sp?-Ferrari) must be one of the most heavily instrumented tracks? Sounds interesting - have you got any further info on any of them? I think dad's library has tons of Jim hall related stuff, but he/we are not really into Prancing horses :) -- Matt Knutsen http://www.knutspeed.com

"Les Neilson" <[email hidden]> skrev i melding news:NcIwa.97019$[email hidden]... > Surely the way GPS works is to move the entire data set by a variableamount > over a period of time, right? So if the ends match up, the accuracybetween > successive points should be to all extents and purposes the maximum > achievable anyway. Yes, but the way we measure relies on a radio base station to get that extreme accuracy. It has a range of quite a few kilometres if the line of sight is ok. Our top plan would be a trolley with four antennas as a minimum that would have telescopic movement for each antenna. Of course - if you want to map every scratch in the surface you would need something even more detailed, maybe laser measurement. But then we're talkin huge $ :) -- Matt Knutsen http://www.knutspeed.com bk AT knutspeed DOT com

For what it's worth I am doing FIA truck racing and we use GPS to to speed control so i have access to quite a lot of tracks and to a GPS box with some memory in it, if anyone is interested I could get a box, put it on a scooter and make a few laps in the late evening to get the coordinates. We have around 2.5 meters accuracy and 1s sample rate but I can make several laps, I was thinking of making one on the inside, middle and outside limits of the tracks so the programmer could get the real layout, by making several laps you use get quite accurate... I'm going next week to Misano, then it'll be A1, then Nogaro, Nurburgring, Alastaro, Most, Zolder, Jarama and Le Mans.

Perhaps not but are you satisfied with the accuracy of current games? "Damien Smith" <[email hidden]> wrote in message news:b9v15f$mtfiv$[email hidden]... > > Sure, but are you telling me that there's a single game out there that has > every corner within 30cm accuracy - please don't say GP4. > >

On 15 May 2003, Txl wrote: ...

Quoted text (7 lines)

> We have around 2.5 meters accuracy and 1s sample rate but I can make > several laps, I was thinking of making one on the inside, middle and > outside limits of the tracks so the programmer could get the real layout, > by making several laps you use get quite accurate... > > I'm going next week to Misano, then it'll be A1, then Nogaro, Nurburgring, > Alastaro, Most, Zolder, Jarama and Le Mans.

With several laps of data, I think you will have many overlapping points. Do you have a scheme for filtering/splining (or ??) to generate one surface that is smooth enough to drive on?

>"Haqsau" <[email hidden]> wrote > Perhaps not but are you satisfied with the accuracy of current games? Absolutely not. Even at Papy, it's often bollocks. Many don't understand the little more effort needed adds so much more atmosphere to the simulation. Only Motorsims with AMA Superbike had barely touched any sense of realistic track accuracy. It's pants for everybody else. Yep even F1RC (but I guess they are helped by the fact that FIA security is so high). -- -- Fran�ois M�nard <ymenard> -- http://ymenard.cjb.net/ -- This announcement is brought to you by the Shimago-Dominguez Corporation - helping America into the New World...

Doug Milliken <[email hidden]> wrote in news:[email hidden]:

Quoted text (15 lines)

> On 15 May 2003, Txl wrote: > ... >> We have around 2.5 meters accuracy and 1s sample rate but I can make >> several laps, I was thinking of making one on the inside, middle and >> outside limits of the tracks so the programmer could get the real >> layout, by making several laps you use get quite accurate... >> >> I'm going next week to Misano, then it'll be A1, then Nogaro, >> Nurburgring, Alastaro, Most, Zolder, Jarama and Le Mans. > > With several laps of data, I think you will have many overlapping > points. Do you have a scheme for filtering/splining (or ??) to > generate one surface that is smooth enough to drive on? > >

I have no "filterring process" I'm waiting for you guys to give me the best method, either make a VERY SLOW laps, like walking speed to get many points close together and then you can correct the possible variations, or make many different lapsat "normal scooter speed" and try to make something out of it...

On 16 May 2003 08:16:34 GMT, Txl <txl@@free.fr> wrote:

Quoted text (22 lines)

>Doug Milliken <[email hidden]> wrote in >news:[email hidden]: > >> On 15 May 2003, Txl wrote: >> ... >>> We have around 2.5 meters accuracy and 1s sample rate but I can make >>> several laps, I was thinking of making one on the inside, middle and >>> outside limits of the tracks so the programmer could get the real >>> layout, by making several laps you use get quite accurate... >>> >>> I'm going next week to Misano, then it'll be A1, then Nogaro, >>> Nurburgring, Alastaro, Most, Zolder, Jarama and Le Mans. >> >> With several laps of data, I think you will have many overlapping >> points. Do you have a scheme for filtering/splining (or ??) to >> generate one surface that is smooth enough to drive on? > >I have no "filterring process" I'm waiting for you guys to give me the >best method, either make a VERY SLOW laps, like walking speed to get many >points close together and then you can correct the possible variations, >or make many different lapsat "normal scooter speed" and try to make >something out of it...

Although I lack time, surely someone can put the data to use. I'd just drive over it at reasonable speed a couple of times; no need to walk endlessly for hours. ;-) You can then pick together the points with perhaps even a 'closest distance' match from all the points. That's how my track editor works when creating a spline patch. Mostly it's no problem. Ruud van Gaal Free car sim: http://www.racer.nl/ Pencil art : http://www.marketgraph.nl/gallery/

[email hidden] (Ruud van Gaal) wrote in news:[email hidden]:

Quoted text (38 lines)

> On 16 May 2003 08:16:34 GMT, Txl <txl@@free.fr> wrote: > >>Doug Milliken <[email hidden]> wrote in >>news:[email hidden]: >> >>> On 15 May 2003, Txl wrote: >>> ... >>>> We have around 2.5 meters accuracy and 1s sample rate but I can >>>> make several laps, I was thinking of making one on the inside, >>>> middle and outside limits of the tracks so the programmer could get >>>> the real layout, by making several laps you use get quite >>>> accurate... >>>> >>>> I'm going next week to Misano, then it'll be A1, then Nogaro, >>>> Nurburgring, Alastaro, Most, Zolder, Jarama and Le Mans. >>> >>> With several laps of data, I think you will have many overlapping >>> points. Do you have a scheme for filtering/splining (or ??) to >>> generate one surface that is smooth enough to drive on? >> >>I have no "filterring process" I'm waiting for you guys to give me the >>best method, either make a VERY SLOW laps, like walking speed to get >>many points close together and then you can correct the possible >>variations, or make many different lapsat "normal scooter speed" and >>try to make something out of it... > > Although I lack time, surely someone can put the data to use. I'd just > drive over it at reasonable speed a couple of times; no need to walk > endlessly for hours. ;-) > You can then pick together the points with perhaps even a 'closest > distance' match from all the points. That's how my track editor works > when creating a spline patch. Mostly it's no problem. > > > Ruud van Gaal > Free car sim: http://www.racer.nl/ > Pencil art : http://www.marketgraph.nl/gallery/ >

Well, i'm going to MISANO this friday and i'll grab a scooter and make some laps friday evening and post the text files to my web page next week, if someone can make something out of it then it's perfect....