I was wrong about importance of g data in driver development
#46
Originally Posted by ColorChange
Brandon:
If you are pushing hard on the entries as I was in the videos, I missed the apexes primarily because of errors on entry (and getting used to a new car). If you asked me to take a reasonable entry speed and back off the trail braking, could I then hit my apexes and exits? The only one who might no that is F1ten. He has seen me and ridden with me at the track. You can't tell that from the videos I have hosted and that's my point.
Secondly, by driving different lines at different g-sums, you can compare segment times (all in the DAS). It really is quite simple.
If you are pushing hard on the entries as I was in the videos, I missed the apexes primarily because of errors on entry (and getting used to a new car). If you asked me to take a reasonable entry speed and back off the trail braking, could I then hit my apexes and exits? The only one who might no that is F1ten. He has seen me and ridden with me at the track. You can't tell that from the videos I have hosted and that's my point.
Secondly, by driving different lines at different g-sums, you can compare segment times (all in the DAS). It really is quite simple.
Simple huh? Gosh I guess we all need to find another HOBBY !
O' Ye of Little Faithe...
Listen to CC...
We've been doing this "all wrong" for decades!!!
#47
Lifetime Rennlist Member
Thread Starter
Originally Posted by ColorChange
Mark:
A former F1 technical director analyzes the best driver in the world using lat and long g analysis, Milliken, Skip Barber, numerous Race Car Engineering articles, numerous SAE technical papers, professional race car consultants/coaches, etc. etc. etc. and you say you have not seen any proof?
errr... aaah ... open your eyes?
You're right on my inability to say the DAS is a better way for others to learn. I am not qualified to make that statement and stand corrected.
How do you know I don't have the line and exit speed reasonably learned (clearly I don't have the entry portion learned well yet)? You may be right by your standards, but I don't think you're qualified to make the statement (lack of information on my driving ability).
A former F1 technical director analyzes the best driver in the world using lat and long g analysis, Milliken, Skip Barber, numerous Race Car Engineering articles, numerous SAE technical papers, professional race car consultants/coaches, etc. etc. etc. and you say you have not seen any proof?
errr... aaah ... open your eyes?
You're right on my inability to say the DAS is a better way for others to learn. I am not qualified to make that statement and stand corrected.
How do you know I don't have the line and exit speed reasonably learned (clearly I don't have the entry portion learned well yet)? You may be right by your standards, but I don't think you're qualified to make the statement (lack of information on my driving ability).
Milliken/Milliken is a little different story as that book is fundamentally about developing a car, not drivers. So again, where do they advocate using g-g graph to develop the driver?
If you can respond quickly, that is great. Otherwise I am out until early next week.
Now, as far as commenting on your driving, you have posted a couple of videos of some laps. This is not meant as an insult - you drive like almost every other novice I have seen once they have a bit of track time and have some confidence in their basic car control. In the videos, I see inconsistent lines, early apexes, little cheats on entry (turn in a bit then finish the turn), late throttle, jerky steering inputs (when I can see the wheel movement) and over-charging the entry to almost every corner. This is the same set of errors I have seen dozens of times with people who have limited track experience. From the videos, I conclude you do not have the line down nor exit speed. How do your lap times compare to SCCA T1, for example. You have as much (more?) suspension that those cars and way better power to weight. They run DOT tires - not sure what you run but that should not be a huge difference. There is a measure of how much of your car you are using.
#48
Lifetime Rennlist Member
Thread Starter
Originally Posted by ColorChange
Secondly, by driving different lines at different g-sums, you can compare segment times (all in the DAS). It really is quite simple.
#49
Mark:
Patience is not my strong suit.
Milliken p 8. “… g-g diagram. By recording the outputs of longitudinal and lateral accelerometers in the vehicle, a plot can be made of driver/vehicle performance.” * * emphasis added
He then goes on to discuss how 3 different drivers utilize the g-g diagram and includes 4 graphs on page 9. He directly uses g-g diagrams to analyze drivers. I don’t think it’s possible to be much clearer than that. But, yes, … I have more.
Pg 348,
“The g-g diagram has two general uses:
* Plotting a particular task performance
* Determine the maneuvering area utilized by the driver/vehicle in a comprehensive series of tasks (i.e. laps on a race course).”
He then goes on to directly analyze drivers performance using g-g plots showing where a poorer driver does not approach the limits and better drivers do. This is developing the driver using g-g plots.
There is more, but this is more than necessary already.
Carroll Smith, Drive To Win
Pg 3-7
Lateral acceleration (lateral g). This one tells us a lot about the performance of both the chassis and the driver. * my emphasis
Pg 3-8
Longitudinal acceleration (longitudinal g). This trace tells us how well the car/driver combination is braking and accelerating.
On this page, Smith recommends using Buddy Fey’s book, Competition Data Logging so we will go to pg 115.
“… the X-Y plot of lateral versus long g makes for interesting study, and can be used to help drivers try to adapt their technique to extract more, if not ultimate, performance from their cars.” * my emphasis
Patience is not my strong suit.
Milliken p 8. “… g-g diagram. By recording the outputs of longitudinal and lateral accelerometers in the vehicle, a plot can be made of driver/vehicle performance.” * * emphasis added
He then goes on to discuss how 3 different drivers utilize the g-g diagram and includes 4 graphs on page 9. He directly uses g-g diagrams to analyze drivers. I don’t think it’s possible to be much clearer than that. But, yes, … I have more.
Pg 348,
“The g-g diagram has two general uses:
* Plotting a particular task performance
* Determine the maneuvering area utilized by the driver/vehicle in a comprehensive series of tasks (i.e. laps on a race course).”
He then goes on to directly analyze drivers performance using g-g plots showing where a poorer driver does not approach the limits and better drivers do. This is developing the driver using g-g plots.
There is more, but this is more than necessary already.
Carroll Smith, Drive To Win
Pg 3-7
Lateral acceleration (lateral g). This one tells us a lot about the performance of both the chassis and the driver. * my emphasis
Pg 3-8
Longitudinal acceleration (longitudinal g). This trace tells us how well the car/driver combination is braking and accelerating.
On this page, Smith recommends using Buddy Fey’s book, Competition Data Logging so we will go to pg 115.
“… the X-Y plot of lateral versus long g makes for interesting study, and can be used to help drivers try to adapt their technique to extract more, if not ultimate, performance from their cars.” * my emphasis
#50
Lifetime Rennlist Member
Thread Starter
I'll go through these with my take...
Milliken/Milliken - Yes, they say what you quoted. You CAN measure the driver/car performance. You left out one key sentence that colors the whole idea "Since we know that these drivers were capable of reaching the vehicle's limit..." Seems to me that Milliken is not discussing how to develop a driver, rather he is discussing how to measure a car by using a world class driver. Where does he describe HOW such data will improve a driver? Measurement, while an important step in improvement, does not guarantee improvement.
Case in point, we can use a ruler to measure our 'attachments' (to steal from Mr Hobbs) but that does nothing to assist us in making them bigger.
I am pretty certain that I could take a world class driver and have them drive a circuit while hugging the inside of each corner and they could max the g's. But they are going to be well off the pace. Later in the book, Milliken uses a g-g graph to show the difference between and amateur and pro. But again, where does he tell us how to use that data to improve. So use my example of running the inside and tell me how g data will tell us there is a problem and how to address it.
Then on to Caroll Smith. Same comments apply, IMHO. He is talking about measuring but where does he tell us how to improve? IF we know the capabilities of the car and the driver is not using it, what do we do and where? And again, if the line is bad but the g's are maxed out, what are we going to see and do?
I will have to give you the one from Buddy Fey. Unfortunately I don't have that book, though I wish I did. Other than that book, what are his credentials? I have heard about that book for a long time but never heard of him outside that environment. You did provide a quote where someone is touting using g data to improve the driver. Now, does he disclose how to do that?
I didn't see anything from Skippy. I have a lot of respect for what he says about driver development - he has a great deal of experience both with his own career and as a teacher.
Milliken/Milliken - Yes, they say what you quoted. You CAN measure the driver/car performance. You left out one key sentence that colors the whole idea "Since we know that these drivers were capable of reaching the vehicle's limit..." Seems to me that Milliken is not discussing how to develop a driver, rather he is discussing how to measure a car by using a world class driver. Where does he describe HOW such data will improve a driver? Measurement, while an important step in improvement, does not guarantee improvement.
Case in point, we can use a ruler to measure our 'attachments' (to steal from Mr Hobbs) but that does nothing to assist us in making them bigger.
I am pretty certain that I could take a world class driver and have them drive a circuit while hugging the inside of each corner and they could max the g's. But they are going to be well off the pace. Later in the book, Milliken uses a g-g graph to show the difference between and amateur and pro. But again, where does he tell us how to use that data to improve. So use my example of running the inside and tell me how g data will tell us there is a problem and how to address it.
Then on to Caroll Smith. Same comments apply, IMHO. He is talking about measuring but where does he tell us how to improve? IF we know the capabilities of the car and the driver is not using it, what do we do and where? And again, if the line is bad but the g's are maxed out, what are we going to see and do?
I will have to give you the one from Buddy Fey. Unfortunately I don't have that book, though I wish I did. Other than that book, what are his credentials? I have heard about that book for a long time but never heard of him outside that environment. You did provide a quote where someone is touting using g data to improve the driver. Now, does he disclose how to do that?
I didn't see anything from Skippy. I have a lot of respect for what he says about driver development - he has a great deal of experience both with his own career and as a teacher.
Last edited by SundayDriver; 09-08-2004 at 09:09 AM.
#51
Originally Posted by SundayDriver
There is the basic problem I have with driving by g's or g-sum. I can mas g-sum on a perfect line, or I can do it on a crappy line. If I am drving at the limit on the crappy line, I am still slow. How can you tell if you are on the fastest line and using all of the car? Segment times! If you have segment times (and look at exit speeds on Type 1 corners) what can you learn from g-sum that you didn't just learn from the times? Seems to me that g-sum is, at best, a predictor of segment/lap times. But why do I need to predict that, after the fact, when I have the times?
#52
Addict
Rennlist Member
Rennlist Member
Join Date: Dec 2001
Location: MA
Posts: 1,010
Likes: 0
Received 0 Likes
on
0 Posts
Originally Posted by SundayDriver
I'll go through these with my take...
Milliken/Milliken - Yes, they say what you quoted. You CAN measure the driver/car performance. You left out one key sentence that colors the whole idea "Since we know that these drivers were capable of reaching the vehicle's limit..." Seems to me that Milliken is not discussing how to develop a driver, rather he is discussing how to measure a car by using a world class driver. Where does he describe HOW such data will improve a driver? Measurement, while an important step in improvement, does not guarantee improvement.
Case in point, we can use a ruler to measure our 'appendages' (to steal from Mr Hobbs) but that does nothing to assist us in making them bigger.
I am pretty certain that I could take a world class driver and have them drive a circuit while hugging the inside of each corner and they could max the g's. But they are going to be well off the pace. Later in the book, Milliken uses a g-g graph to show the difference between and amateur and pro. But again, where does he tell us how to use that data to improve. So use my example of running the inside and tell me how g data will tell us there is a problem and how to address it.
Milliken/Milliken - Yes, they say what you quoted. You CAN measure the driver/car performance. You left out one key sentence that colors the whole idea "Since we know that these drivers were capable of reaching the vehicle's limit..." Seems to me that Milliken is not discussing how to develop a driver, rather he is discussing how to measure a car by using a world class driver. Where does he describe HOW such data will improve a driver? Measurement, while an important step in improvement, does not guarantee improvement.
Case in point, we can use a ruler to measure our 'appendages' (to steal from Mr Hobbs) but that does nothing to assist us in making them bigger.
I am pretty certain that I could take a world class driver and have them drive a circuit while hugging the inside of each corner and they could max the g's. But they are going to be well off the pace. Later in the book, Milliken uses a g-g graph to show the difference between and amateur and pro. But again, where does he tell us how to use that data to improve. So use my example of running the inside and tell me how g data will tell us there is a problem and how to address it.
This stuff is the end game not the starting point. Driving via tach is not an easily acquired skill. Not to accuse, CC, but is it that you're avoiding it for at the moment its just too difficult to do as you're still struggling with sensory overload. I know that after two years, I'm just beginning to have enough free brains cells to actually look at the tach, let alone note RPM at the time.
You mentioned Skip Barber. Every year we have an instructors/advanced drivers meeting with a lecture given by one of their top guys. In fact, last year the lecturer was extremely enlightening as to his approach to a new track. He described his first trip to the Nurembergring. With a few laps, he was running appropriate fast lap times. How he do it? No DAS. He (and I admit this is amazing) remembered each of the important corners, tried a line, noted his tach speed. Within a half dozen or so laps of trying line and noting tach he was there. BTW, never has anyone from Barber touted DAS as a fundemental teaching tool at the lectures I've attended.
There's no doubt some benefit to gathering as much data as possible, but at least at my level, it is a condiment, not the meal. For the moment, I being eating my steak raw and plain, hold the A1.
#53
Thanks 944turbos. I'm glad you get it. I just wish I was more effective at convincing others.
Mark:
I’m sorry but I give up. I have shown from numerous sources, from Milliken, to F1 technical directors, to the only text dedicated to race car data analysis, to many others, that they use g analysis, and even specifically g-g plots to analyze driver performance. Maybe what you are missing is that if you are not near the fc, you should go faster or change your line.
For the others I will state this as plainly as I can. There are many ways to improve driver performance; one particularly effective way is by using g analysis in combination with DAS driver inputs.
What I prefer to do (and have learned and been taught by all the references I have previously listed) is to plot g-sum as a function of distance. Any time I drop off the limits of my car (that are clearly and easily, reasonably knowable) I try to investigate why. If it is a track condition (camber change, elevation change, …), that should be repeatable. If it is not, I will look at the driver inputs: did I turn the wheel, ease off of the throttle, etc. An additional aid to driving near the limit is to plot the derivative of g-sum or “smoothness” as a function of distance and see how smoothly I am able to achieve a high g-sum.
Once I can drive near the limit consistently (maximize g-sum as a function of distance), I will start to work on the ideal lines. I will do this by driving different lines through turns at consistently high g-sums (much like a test car driver does) and then compare segment times to see which is fastest. (Aim is close to releasing a differential GPS system that will allow you to quite accurately plot and analyze the line you drove http://www.aim-sportline.com/ click on auto, then MXL to see a picture. I can’t wait for that.)
This allows you to do what the fastest driver’s job is, maximize g’s in the appropriate direction. I translate that to say maximize g-sum on the ideal line. I have shown how to break that problem down into first, maximize g-sum, and then segment time test various lines that are driven near the limit to identify the ideal line.
Mark:
I’m sorry but I give up. I have shown from numerous sources, from Milliken, to F1 technical directors, to the only text dedicated to race car data analysis, to many others, that they use g analysis, and even specifically g-g plots to analyze driver performance. Maybe what you are missing is that if you are not near the fc, you should go faster or change your line.
For the others I will state this as plainly as I can. There are many ways to improve driver performance; one particularly effective way is by using g analysis in combination with DAS driver inputs.
What I prefer to do (and have learned and been taught by all the references I have previously listed) is to plot g-sum as a function of distance. Any time I drop off the limits of my car (that are clearly and easily, reasonably knowable) I try to investigate why. If it is a track condition (camber change, elevation change, …), that should be repeatable. If it is not, I will look at the driver inputs: did I turn the wheel, ease off of the throttle, etc. An additional aid to driving near the limit is to plot the derivative of g-sum or “smoothness” as a function of distance and see how smoothly I am able to achieve a high g-sum.
Once I can drive near the limit consistently (maximize g-sum as a function of distance), I will start to work on the ideal lines. I will do this by driving different lines through turns at consistently high g-sums (much like a test car driver does) and then compare segment times to see which is fastest. (Aim is close to releasing a differential GPS system that will allow you to quite accurately plot and analyze the line you drove http://www.aim-sportline.com/ click on auto, then MXL to see a picture. I can’t wait for that.)
This allows you to do what the fastest driver’s job is, maximize g’s in the appropriate direction. I translate that to say maximize g-sum on the ideal line. I have shown how to break that problem down into first, maximize g-sum, and then segment time test various lines that are driven near the limit to identify the ideal line.
Last edited by ColorChange; 09-08-2004 at 09:42 AM.
#54
Lifetime Rennlist Member
Thread Starter
Originally Posted by 944TURBOS
actually color change is right. Maximizing G-sums on different llines is a great tool. Sure you might not be fastest on the "other" lines, but you know that in racing you arent always taking the optimal line because there are other people taking that same line and if you were to keep doing the same thing as them there would be no way to pass. Therefore you have to play around with different lines. yes you might be slower in some aspects but you may be gaining positions. The DAS will show you how to make the best of all the possibly lines. I think CC is misunderstood on this forum. I misunderstood him too until I worked with a DAS and learned from coaches
I am also curious how the g based coaching compared to speed and laptime based coaching. Have you gotten other coaching before using DAS? Do you use the speed data (do you have speed data)?
Thanks
#55
Lifetime Rennlist Member
Thread Starter
Originally Posted by ColorChange
Mark:
I’m sorry but I give up. I have shown from numerous sources, from Milliken, to F1 technical directors, to the only text dedicated to race car data analysis, to many others, that they use g analysis, and even specifically g-g plots to analyze driver performance. Maybe what you are missing is that if you are not near the fc, you should go faster or change your line.
I’m sorry but I give up. I have shown from numerous sources, from Milliken, to F1 technical directors, to the only text dedicated to race car data analysis, to many others, that they use g analysis, and even specifically g-g plots to analyze driver performance. Maybe what you are missing is that if you are not near the fc, you should go faster or change your line.
#57
Once I can drive near the limit consistently (maximize g-sum as a function of distance), I will start to work on the ideal lines.
#58
Burning Brakes
Join Date: Jul 2003
Location: Cape Cod
Posts: 856
Likes: 0
Received 0 Likes
on
0 Posts
It's just like learning to play golf by watching comparisons of your swing with Tiger's, or Jack's or whomever! All you have to do is change your swing to match theirs! Or follow the data points on a computer print out!
Then go out onto any course and perfectly replicate, in changing conditions (lie, wind, humidity), what you have learned. For any shot there will be an optimum swing.....just do it!!! Just like for any corner, at any time, on any track there will be an optimum line for any given car! It will never be exactly the same! ( temperature, tire wear, GVW, etc) All you have to do is find it!
Data points will always help, but so will practice, AND...... experience, even when gleaned from an instructor or other drivers expertise. The synthesis of all available inputs would seem to allow the quickest optimization for all situations!
Then go out onto any course and perfectly replicate, in changing conditions (lie, wind, humidity), what you have learned. For any shot there will be an optimum swing.....just do it!!! Just like for any corner, at any time, on any track there will be an optimum line for any given car! It will never be exactly the same! ( temperature, tire wear, GVW, etc) All you have to do is find it!
Data points will always help, but so will practice, AND...... experience, even when gleaned from an instructor or other drivers expertise. The synthesis of all available inputs would seem to allow the quickest optimization for all situations!
#59
CC...
Let's start with how I see the basics. Yes, I agree that you can look at G-sum data and evaluate a driver (hey now, let me explain). When we look at top-notch drivers, we expect them to be driving the line well. We expect them to be on the edge of the limit as much as possible. What the G-sum graph is showing is how high they were able to make the limit through a combination of car setup and driver skill. You can take two drivers and each might have a different "limit" depending on their style of driving. The G-sum data can show us that a certain driver is doing something different that creates a slightly higher limit.
On the flip side, having a high G-sum doesn't necessarily mean you are a good driver. You can be on a slow line and driving it as fast as possible. For example, you could overbrake a turn (high G's), and then trail brake into a very tight line, which gives you great lat G's. Finally when you start nailing the gas, if you keep the car pinched in, you will still have good G's. I think it's immensely obvious that you can create high G-sums while still being slow.
I think the greatest use you'll get out of the DAS is letting a pro drive your car and then overlaying their DAS data on top of yours. I bet you'll be shocked by how much higher their limits are than yours. Even with begging and pleading, I doubt you'll find many PCA instructors who will drive your car near the limits as few want to be on the hook for a 6-digit screwup. Again, where is GhettoRacer when we really need him???
Let's start with how I see the basics. Yes, I agree that you can look at G-sum data and evaluate a driver (hey now, let me explain). When we look at top-notch drivers, we expect them to be driving the line well. We expect them to be on the edge of the limit as much as possible. What the G-sum graph is showing is how high they were able to make the limit through a combination of car setup and driver skill. You can take two drivers and each might have a different "limit" depending on their style of driving. The G-sum data can show us that a certain driver is doing something different that creates a slightly higher limit.
On the flip side, having a high G-sum doesn't necessarily mean you are a good driver. You can be on a slow line and driving it as fast as possible. For example, you could overbrake a turn (high G's), and then trail brake into a very tight line, which gives you great lat G's. Finally when you start nailing the gas, if you keep the car pinched in, you will still have good G's. I think it's immensely obvious that you can create high G-sums while still being slow.
I think the greatest use you'll get out of the DAS is letting a pro drive your car and then overlaying their DAS data on top of yours. I bet you'll be shocked by how much higher their limits are than yours. Even with begging and pleading, I doubt you'll find many PCA instructors who will drive your car near the limits as few want to be on the hook for a 6-digit screwup. Again, where is GhettoRacer when we really need him???
#60
Race Director
BrianP and I autocrossed his Boxster S this past weekend. He has a DL90 hooked up.
Brian: could you post our G-sum graphs from the fastest runs we both made? (Mine was the last run of the day - the last set of data on the box) I only recorded my last two runs, so datasets 1 through 8 are yours, and datasets 9, 10 are mine.
Let's not tell the folks who did better, and see if that can be determined by just looking at the plots!
I am curious to see how our plots compare.
Brian's 986S has a modified suspension (more suited for track than autox), but we were running on street tires. Our times were very close, btw.
Perhaps this may help dispell or verify some theories being thrown around here. Emperical data baby! Gotta love hard evidence.
-Z-man.
Brian: could you post our G-sum graphs from the fastest runs we both made? (Mine was the last run of the day - the last set of data on the box) I only recorded my last two runs, so datasets 1 through 8 are yours, and datasets 9, 10 are mine.
Let's not tell the folks who did better, and see if that can be determined by just looking at the plots!
I am curious to see how our plots compare.
Brian's 986S has a modified suspension (more suited for track than autox), but we were running on street tires. Our times were very close, btw.
Perhaps this may help dispell or verify some theories being thrown around here. Emperical data baby! Gotta love hard evidence.
-Z-man.