DL1 datalogger questions
03-03-2005, 02:41 AM
#16
Instructor
Thread Starter
Join Date: Apr 2002
Location: Caledon East Ontario
Posts: 159
Likes: 0
Received 0 Likes
on
0 Posts
yeah, I posted on honda tech as well, trying to get as much input as possible.
I agree with Mark that the video is the most important, the GPS allows for multiple segment times, and vehicle speed which I feel would be useful, but the track map, and line won't provide much useful data.
I am going to go ahead and order up a system tomorrow, I called the one canadian dealer today but received no answer.
Powerworks sounds good, board sponsor, everyone had positive dealings with them, any other suggestions before I go ahead?
Thanks,
Andrew
I agree with Mark that the video is the most important, the GPS allows for multiple segment times, and vehicle speed which I feel would be useful, but the track map, and line won't provide much useful data.
I am going to go ahead and order up a system tomorrow, I called the one canadian dealer today but received no answer.
Powerworks sounds good, board sponsor, everyone had positive dealings with them, any other suggestions before I go ahead?
Thanks,
Andrew
03-03-2005, 09:48 AM
#17
It seems clear that the two major systems of calculating track position are less than perfect.
The 'traditional' approach relies on computing a position based on speed and G factors, which suffers from accuracy and repeatability of the input sensors. Accelerometers assume that the platform is flat and stable, which is clearly not the case in a race car. The net result is compounding error creep over time.
The GPS approach is subject to small but real repeatability errors due to the fact that the grid of satellites used for a given position fix changes over time due to reception and weather factors.
Neither system is perfect. As Mark says, the ability to see the DACs data [RPM, Throttle, Brake, G loading, speed] overlayed on a video which shows position on track, as well as traffic/obstructions/etc is the key to post-run analysis.
The video has another less widely used advantage in qualifying and test as well as DE applications. AS the driver, you can record a commentary of what you are doing/trying during the run. Most cameras will pick up the drivers voice [some might have to shout!], and this is a great method to figure out what really works. When you can see & hear the video AND see the data together, its quickly become a terrific tool for showing what works best.
The 'traditional' approach relies on computing a position based on speed and G factors, which suffers from accuracy and repeatability of the input sensors. Accelerometers assume that the platform is flat and stable, which is clearly not the case in a race car. The net result is compounding error creep over time.
The GPS approach is subject to small but real repeatability errors due to the fact that the grid of satellites used for a given position fix changes over time due to reception and weather factors.
Neither system is perfect. As Mark says, the ability to see the DACs data [RPM, Throttle, Brake, G loading, speed] overlayed on a video which shows position on track, as well as traffic/obstructions/etc is the key to post-run analysis.
The video has another less widely used advantage in qualifying and test as well as DE applications. AS the driver, you can record a commentary of what you are doing/trying during the run. Most cameras will pick up the drivers voice [some might have to shout!], and this is a great method to figure out what really works. When you can see & hear the video AND see the data together, its quickly become a terrific tool for showing what works best.
03-03-2005, 11:21 AM
#18
Advanced
Join Date: Aug 2003
Location: Norway
Posts: 93
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by James Achard
Regarding the accuracy the quote below is from the Race Technology website.
"With good GPS reception, positional accuracy is about 3m (CEP)."
It seems to be a bit better in my initial findings but this was at Sebring where there is no elevation change and ideal conditions.
Cheers, James
"With good GPS reception, positional accuracy is about 3m (CEP)."
It seems to be a bit better in my initial findings but this was at Sebring where there is no elevation change and ideal conditions.
Cheers, James
/Ĝrjan
03-03-2005, 11:25 AM
#19
Advanced
Join Date: Aug 2003
Location: Norway
Posts: 93
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by APKhaos
<some snipped>
Accelerometers assume that the platform is flat and stable, which is clearly not the case in a race car. The net result is compounding error creep over time.
Accelerometers assume that the platform is flat and stable, which is clearly not the case in a race car. The net result is compounding error creep over time.
/Ĝrjan
03-03-2005, 11:38 AM
#20
Cruisin'
Join Date: Dec 2004
Posts: 9
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by forklift
LOL, Hey Ryan, is that you? Thanks for the ride into the grass in T4!
Seriously, great driving though, you were fast on track on Sunday....
Jim
Seriously, great driving though, you were fast on track on Sunday....
Jim
Your were pretty fast yourself..
The real question I would have (never having used GPS data for racing) is whether you can interpret the data to tell you something meaningful. So you were 1 foot different from one lap to another and the first was a faster segment time. Now what did you do and how do you repeat it. I am not sure the GPS data would let you understand that. That is where I think video does the most for a driver.
Example track map..Not the best ones tho
Beaver Run Track Map and G-G Circle
03-03-2005, 11:41 AM
#21
Sipmlest way to understand this is to imagine a 1G loading perfectly aligned with the accelerometer's axis. This will result in a 1G signal from the device +/- errors due to temperature/calibration etc.
Now imagine a 1G load at 45 degrees to the accelerometer's axix - the kind of thing that happens at the apex of a turn when you hit the gator, for example. This will result in a >1G signal from the device. It does not read the load vector that is off its measurement axis.
I'm not sure if RT attempts to correct GPS data using accelerometer signals, but I doubt it. I recall that RT's code does some smoothing of the GPS data, but it seems to me that using the accelerometer data would introduce more error than it would correct. I could be wrong, but best bet is to confirm with RT directly.
Now imagine a 1G load at 45 degrees to the accelerometer's axix - the kind of thing that happens at the apex of a turn when you hit the gator, for example. This will result in a >1G signal from the device. It does not read the load vector that is off its measurement axis.
I'm not sure if RT attempts to correct GPS data using accelerometer signals, but I doubt it. I recall that RT's code does some smoothing of the GPS data, but it seems to me that using the accelerometer data would introduce more error than it would correct. I could be wrong, but best bet is to confirm with RT directly.
03-03-2005, 03:02 PM
#22
Addict
Rennlist
Lifetime Member
Rennlist
Lifetime Member
Interesting discussion. I'm not familiar with commercial GPS applications. How many satellites are you using for position fix and what kind of filtering? 3M CEP sounds extremely optimistic, and expect that would surely be relatvie accuracy as discussed earlier.
03-04-2005, 05:44 AM
#24
3rd Gear
Join Date: Mar 2005
Posts: 3
Likes: 0
Received 0 Likes
on
0 Posts
Hi Everyone,
My name is Andy; I'm the software/GPS guy from Race Technology whom manufactures the DL1/AX22 products. Ĝrjan invited me to join in with the discussion, I'm not here to advertise honest(!)... just answer a couple of the points that have been raised.
First GPS accuracy, I could write a book on the subject as there is lots to it, but briefly....
Accuracy
We quote 3m CEP, which is about right for a typical tree lined circuit with buildings blocking clear views to the horizons. However, at a clear track (e.g. and airfield) then the accuracy comes down to well under a meter, conversely with poor GPS conditions, i.e. surrounded by buildings it can also go up. We choose to be conservative with the specifications and quote 3m CEP as it much better to get people saying it better than expected, than visa-versa! There is lots of rubbish quoted about GPS accuracy, 3m CEP is as good as it gets for a standalone GPS receiver without taking about a subscription to the very expensive differential services ($k's a year), or having a trackside differential correction system. The 3m error is limited by the physics of the GPS problem, there is very little error added by the GPS receiver itself. Most GPS manufacturers quote accuracies under ideal conditions and these can be misleading to say the least! Also: never confuse accuracy and resolution as the 2 are unrelated, our GPS system has a resolution of 0.1mm!
Causes of GPS errors
With a clear view of the sky the GPS positional errors are caused by atmospheric phenomena and inaccuracies in the satellite orbits these problems typically cause the GPS position to wander around quite slowly, over the period of 10s of minutes (its not seconds, but its not hours either). With buildings/trees in close proximity, positional errors are also induced from the reflected GPS radio signals (AKA multi-path), this can cause quite fast positional errors. Multi-path is a big problem when using you sat-nav system around town, but fortunately they dont normally build skyscrapers by race tracks!
Combining accelerometers and GPS
Combining accelerometers and GPS does not normally improve the overall positional accuracy a great deal (in practical terms we get about 20% improvement), however it does allow use to every effectively "filter the GPS data to fill in any dropouts and eliminate and large, short term GPS errors it basically cleans the data up in an intelligent way.
Comparison of GPS and lap beacon wheel speed systems.
Optical lap beacons have a beam spread of about 2-3m (radio transponders are about the same). So if you have good/average GPS conditions then GPS based lap timing is a bit more accurate, with poor GPS coverage then an optical lap beacon is a bit more accurate in practice neither can promise 0.01s accuracy. Assuming your on doing 100mph on the straight then 2m takes about 0.04s. Line up 10 optical systems at the side of the track and 10 lap timers in the car and I can promise that is about the spread you will see (weve done it!). But that isnt the end of the story . A lap beacon system only knows where you are once per lap, so you get a valid lap time but the intermediate sector times are based on the highly inaccurate wheel speeds so they are basically all over the place! In contrast GPS knows where you are the whole time so the lap and sector times are all accurate. There is also the practical problems of using optical beacons to consider ..
Speed errors
Speed and position are calculated almost independently the accuracy of speed is a completely different topic to position. Errors in speed are better defined than position typically speed errors are in the range 0.05mph to 0.3mph. In contrast to position these errors are mainly due to errors in the GPS receiver itself (its the carrier noise), the receiver in the DL1/AX22 has a carrier noise around 0.6mm which is the currently as good as it gets using todays semiconductors. Ive read on other manufacturers websites speed accuracies of 0.01mph from GPS, Ive no idea where they get this figure from but it is currently fundamentally impossible regardless of cost! The speed error is so low, the error can normally be neglected altogether.
And now a quick word on the video overlay stuff .
The new v5.1 is on the web site which is a significant improvement over v5.0 in terms of the video export functionality. If anyone would like to make suggestions about how we could improve things further then now would be a good time to suggest it we are well stuck into v6.0 of the software and welcome all feedback!
Blimey thats enough typing for now, better get back to work ;o)
My name is Andy; I'm the software/GPS guy from Race Technology whom manufactures the DL1/AX22 products. Ĝrjan invited me to join in with the discussion, I'm not here to advertise honest(!)... just answer a couple of the points that have been raised.
First GPS accuracy, I could write a book on the subject as there is lots to it, but briefly....
Accuracy
We quote 3m CEP, which is about right for a typical tree lined circuit with buildings blocking clear views to the horizons. However, at a clear track (e.g. and airfield) then the accuracy comes down to well under a meter, conversely with poor GPS conditions, i.e. surrounded by buildings it can also go up. We choose to be conservative with the specifications and quote 3m CEP as it much better to get people saying it better than expected, than visa-versa! There is lots of rubbish quoted about GPS accuracy, 3m CEP is as good as it gets for a standalone GPS receiver without taking about a subscription to the very expensive differential services ($k's a year), or having a trackside differential correction system. The 3m error is limited by the physics of the GPS problem, there is very little error added by the GPS receiver itself. Most GPS manufacturers quote accuracies under ideal conditions and these can be misleading to say the least! Also: never confuse accuracy and resolution as the 2 are unrelated, our GPS system has a resolution of 0.1mm!
Causes of GPS errors
With a clear view of the sky the GPS positional errors are caused by atmospheric phenomena and inaccuracies in the satellite orbits these problems typically cause the GPS position to wander around quite slowly, over the period of 10s of minutes (its not seconds, but its not hours either). With buildings/trees in close proximity, positional errors are also induced from the reflected GPS radio signals (AKA multi-path), this can cause quite fast positional errors. Multi-path is a big problem when using you sat-nav system around town, but fortunately they dont normally build skyscrapers by race tracks!
Combining accelerometers and GPS
Combining accelerometers and GPS does not normally improve the overall positional accuracy a great deal (in practical terms we get about 20% improvement), however it does allow use to every effectively "filter the GPS data to fill in any dropouts and eliminate and large, short term GPS errors it basically cleans the data up in an intelligent way.
Comparison of GPS and lap beacon wheel speed systems.
Optical lap beacons have a beam spread of about 2-3m (radio transponders are about the same). So if you have good/average GPS conditions then GPS based lap timing is a bit more accurate, with poor GPS coverage then an optical lap beacon is a bit more accurate in practice neither can promise 0.01s accuracy. Assuming your on doing 100mph on the straight then 2m takes about 0.04s. Line up 10 optical systems at the side of the track and 10 lap timers in the car and I can promise that is about the spread you will see (weve done it!). But that isnt the end of the story . A lap beacon system only knows where you are once per lap, so you get a valid lap time but the intermediate sector times are based on the highly inaccurate wheel speeds so they are basically all over the place! In contrast GPS knows where you are the whole time so the lap and sector times are all accurate. There is also the practical problems of using optical beacons to consider ..
Speed errors
Speed and position are calculated almost independently the accuracy of speed is a completely different topic to position. Errors in speed are better defined than position typically speed errors are in the range 0.05mph to 0.3mph. In contrast to position these errors are mainly due to errors in the GPS receiver itself (its the carrier noise), the receiver in the DL1/AX22 has a carrier noise around 0.6mm which is the currently as good as it gets using todays semiconductors. Ive read on other manufacturers websites speed accuracies of 0.01mph from GPS, Ive no idea where they get this figure from but it is currently fundamentally impossible regardless of cost! The speed error is so low, the error can normally be neglected altogether.
And now a quick word on the video overlay stuff .
The new v5.1 is on the web site which is a significant improvement over v5.0 in terms of the video export functionality. If anyone would like to make suggestions about how we could improve things further then now would be a good time to suggest it we are well stuck into v6.0 of the software and welcome all feedback!
Blimey thats enough typing for now, better get back to work ;o)
03-04-2005, 06:08 AM
#25
Advanced
Join Date: Aug 2003
Location: Norway
Posts: 93
Likes: 0
Received 0 Likes
on
0 Posts
Thanks Andy - I'm most certain that your presence and expertice are welcome at these forums. This is definately not the first time your products have been discussed. Myself, I just try not to post anything misleading 
/Ĝrjan
/Ĝrjan
03-04-2005, 06:20 PM
#26
Rennlist Member
COOL- Somebody from RT joined the forum! Welcome to the board, Andy. And thanks for the explanation.
I used the DL1 last summer and agree with the above comments that you can in fact see the lines you take. But I didn't find this to be helpful; i mean, what do you do with that? I know what line I should be taking, and that often I miss it, causing slow exit speeds. So I don't think analyzing my lines helps that any. There's plenty other stuff to look at. My latest fascination is the Time Slip graphs. Uber-cool.
At any rate, accuracy just isn't a big deal to me. It's accurate. Nuff said. It's a legit question, though.
I'd love to hear comments on video software (RT versus Trackvision) since I just bought a digital camcorder. In looking at the RT software (http://www.race-technology.com/WebPa...strations.html) it doesn't appear that you can customize the virtual dashboard. That's bad bad news if that's true. Maybe Trackvision is the way to go...
I used the DL1 last summer and agree with the above comments that you can in fact see the lines you take. But I didn't find this to be helpful; i mean, what do you do with that? I know what line I should be taking, and that often I miss it, causing slow exit speeds. So I don't think analyzing my lines helps that any. There's plenty other stuff to look at. My latest fascination is the Time Slip graphs. Uber-cool.
At any rate, accuracy just isn't a big deal to me. It's accurate. Nuff said. It's a legit question, though.
I'd love to hear comments on video software (RT versus Trackvision) since I just bought a digital camcorder. In looking at the RT software (http://www.race-technology.com/WebPa...strations.html) it doesn't appear that you can customize the virtual dashboard. That's bad bad news if that's true. Maybe Trackvision is the way to go...
03-04-2005, 06:53 PM
#27
3rd Gear
Join Date: Mar 2005
Posts: 3
Likes: 0
Received 0 Likes
on
0 Posts
Video overlay
Hi,
Well, since I wrote the RT video software I'm not sure I'm the ideal person to give an impartial opinion!.. however I can confirm that you can define your own dashboards, the 6 dashboard layouts that come with the installation are just examples. The dashboards are defined using a simple script language which is defined in the help files - so you can make your own up or modify there as required.
I have seen trackvision in action and it also looks really cool... and they are a good bunch of guys, whom offer good support and know their stuff. I guess if there is an advantage of the RT system it is that it is fully integrated into the analysis software and under constant development so you can be sure that it will only improve over the months and years, I'm deep into version 6 and it already looks fantastic - it's the biggest new release by some way... however I'm not a salesman, just a techie... that's is enough sales-stuff from me! Just technical questions please ;o)
Andy
Well, since I wrote the RT video software I'm not sure I'm the ideal person to give an impartial opinion!.. however I can confirm that you can define your own dashboards, the 6 dashboard layouts that come with the installation are just examples. The dashboards are defined using a simple script language which is defined in the help files - so you can make your own up or modify there as required.
I have seen trackvision in action and it also looks really cool... and they are a good bunch of guys, whom offer good support and know their stuff. I guess if there is an advantage of the RT system it is that it is fully integrated into the analysis software and under constant development so you can be sure that it will only improve over the months and years, I'm deep into version 6 and it already looks fantastic - it's the biggest new release by some way... however I'm not a salesman, just a techie... that's is enough sales-stuff from me! Just technical questions please ;o)
Andy
03-04-2005, 07:05 PM
#28
Great to have Andy here on Rennlist. He has developed a superb product in the DL series. You guys should take a look at his new dash display products - they bring the Race Tech loggers onto the real time display and log market with MoTeC, AIM and others. Very cool.
Guess we have been talking to Andy on and off for well over a year, and we had the pleasure of showing each other our respective stuff at PRI last November. RT build great product, which I can say with confidence even though our TrackVision product happens to compete with RT in one narrow product area.
Dammit, I nearly said why we are better at that bit, but that would have been cheap.
Guess we have been talking to Andy on and off for well over a year, and we had the pleasure of showing each other our respective stuff at PRI last November. RT build great product, which I can say with confidence even though our TrackVision product happens to compete with RT in one narrow product area.
Dammit, I nearly said why we are better at that bit, but that would have been cheap.
03-04-2005, 07:29 PM
#29
Burning Brakes
Join Date: Nov 2001
Location: Jupiter
Posts: 1,093
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by Eric86Red911
I used the DL1 last summer and agree with the above comments that you can in fact see the lines you take. But I didn't find this to be helpful; i mean, what do you do with that? I know what line I should be taking, and that often I miss it, causing slow exit speeds. So I don't think analyzing my lines helps that any.
Stephen
