Apex Digital Track Coach Review
#16
Nordschleife Master
Thread Starter
Thanks Joe, I was looking forward to your write up!
All I got was this part
"if you are driving near 10/10ths, you aren't going to be look at lights."
And that lap in the GT3 is super sexy driving.
Is that your PB? that's the one I'm going to be watching on the plane ride over
All I got was this part
"if you are driving near 10/10ths, you aren't going to be look at lights."
And that lap in the GT3 is super sexy driving.
Is that your PB? that's the one I'm going to be watching on the plane ride over
#17
Rennlist Member
Pffff....
BTW, apprearantly I drive 9.5/10th compared to that lap.
I just looked at the lap again a couple times and funny how the track comes back to me after a few years, even though my laps were 1.53's OMG..
Man, I would have paid to be in the car.
I mean because in 997RS I drive that "loose" but I haven't been able to in RS.. you are doing exactly what I would do if I knew the track well enough...and it's hard in a 991!!
My hats off to you drover.
BTW, apprearantly I drive 9.5/10th compared to that lap.
I just looked at the lap again a couple times and funny how the track comes back to me after a few years, even though my laps were 1.53's OMG..
Man, I would have paid to be in the car.
I mean because in 997RS I drive that "loose" but I haven't been able to in RS.. you are doing exactly what I would do if I knew the track well enough...and it's hard in a 991!!
My hats off to you drover.
#18
Nordschleife Master
Thread Starter
Pffff....
BTW, apprearantly I drive 9.5/10th compared to that lap.
I just looked at the lap again a couple times and funny how the track comes back to me after a few years..
Man, I would have paid to be in the car.
I mean because in 997RS I drive that "loose" but I haven't been able to in RS.. you are doing exactly what I would do if I knew the track well enough...and it's hard in a 991!!
My hats off to you drover.
BTW, apprearantly I drive 9.5/10th compared to that lap.
I just looked at the lap again a couple times and funny how the track comes back to me after a few years..
Man, I would have paid to be in the car.
I mean because in 997RS I drive that "loose" but I haven't been able to in RS.. you are doing exactly what I would do if I knew the track well enough...and it's hard in a 991!!
My hats off to you drover.
Man, you're gonna be kicking *** before the end of your west coast stint coming up, no doubt! Looking forward to seeing you again soon, bro!
#21
Hey everyone, awesome questions. I'll do my best to answer as many as I can!
@kfmcmahon, I'm an instructor at PSDS, so hopefully we'll be using the device in the school before too long. There's a big difference between selling to small businesses, and selling to corporate entities. The main difference is time! But the discussion here and growth of our customer base will be the best way to get APEX at PSDS soon.
Here was my answer to a question from Brian Lee that seems to cover a lot of what is being asked above:
Q: I'm curious about how it is able to determine what the limits are of my car at a given point in time, on a specific track with specific track conditions? I'm guessing part of it has to do with calibrating the device. But lets say while calibrating the device, I purposely take it easy e.g. driving 3/10th's of my ability perhaps due to traffic or just because its my first session of the day. Then later on, I start driving closer 7/10th's of my ability. When I start driving closer to 7/10th's I'd expect to see more green LEDs and less red. But that isn't really telling me I'm getting closer to the car's limit, correct?
A: The "calibration" you are referring to happens constantly. That's why it's considered artificial intelligence. It will constantly learn what the "limit" is based off of your driving. So if you're at 3/10 on average for a lap or so, it will think the limit is around 5/10, or whatever it has determined that "limit" to be based off of that lap, but as soon as you push it closer to 7-8/10th's, even in one corner, it has tons of new information about the limit that can be applied to every situation the car is in.
It samples over 12,000 data points/second, so any minor performance change is detected very quickly. If you're only able to drive the car at 5-6/10th's then the device may only be able to understand the limit to be 7-8/10th's. But, if you carry a lot of speed in one corner (have to enter off-line maybe, or accidentally brake too late or too little), and approach the true limit of the tire (understeer or oversteer) APEX will learn the true limit with shocking accuracy.
Through my professional coaching exploits I've noticed that most drivers, especially in cars with street tires, will approach the limit at least once during a 20-30 minute session. The limit may be found under braking, like getting into the ABS, or during acceleration (PSM/traction control intervention), but as soon as the device gets an idea of that limit, it can more accurately predict the "limit" everywhere else!
So all it takes is one "moment" and the device learns what the tire can take, and can then extrapolate that into other scenarios it's experiences around the track.
It does take at least half a lap at decent pace to build a useful simulation model for the LED display to be relevant. After 1-2 laps the feedback will be relevant to the driver. Similarly, one quick autocross run would be enough to build a robust simulation as well.
In reference to the question about tire degradation etc.. There is some "backwards learning" built into the model, but the best way to get an idea of tire deg, track temp, rain or any other grip condition that may decrease grip as the day wears on is the APEX score that is calculated in the post session data for each lap. Monitor your morning (cool track temps, fresh tires/brakes) score, vs. your afternoon score. You might find you're driving the car the same way but not quite able to achieve the same peak score as you were in cooler temps. That's showing you that the amount of grip available has decreased overall.
Joe, the reason you don't always see the LED bar full of lights is because when the LED bar is 100% full that's representing that 100% grip is available (i.e. the most/peak grip that it thinks is available that day, at that track etc..). This is likely only in a high grip scenario, a corner with good camber, or a different pavement gradient than elsewhere on the lap that provides more grip, or somewhere that your car really excels. If the device is on Beginner or Intermediate Skill Level the resolution is lower (meaning it's showing 0-100% on beginner and 75%-100% on intermediate) so you may see the LED bar full more often. If we don't see the full LED bar illuminated it means that the peak grip level cannot be achieved because either you're on an off-camber corner, on a rise or peak of some sort (braking into the corkscrew at Laguna for example), or on a low-grip surface (turns 3-4 at Laguna) etc..
@randr: Do most people watch their G-meter as they go around the track at speed? I know I don't and I've driven plenty of cars that have them. APEX is different in a few ways:
1) Calculates a lot more than just G's. It's building a friction circle model of your vehicle as you drive, and uses a 9 axis IMU, 10hz gps, accelerometer and gyro to do so. It also will interpolate the limit. The more driving you do around the 7-8/10ths mark, the closer it can get to predicting what 10/10ths actually is. It doesn't have to see the actual limit to interpolate what that limit is.
2) The display is designed to be used similarly to shift lights. Once you adapt to it being there it becomes an easily understood gauge, not one that requires your direct vision to interpret. Although a G-meter may be able to be understood by some, our goal is for the APEX display to intuitive to most, and I think the chasing LED display provides that.
@Manifold, although it is building a model of the car's potential it can only learn from the performance that you show it. So If you never get anywhere near the limit, then it is essentially showing you your limit. Does that make sense?
Thanks for the questions folks, keep them coming!
Andrew
@kfmcmahon, I'm an instructor at PSDS, so hopefully we'll be using the device in the school before too long. There's a big difference between selling to small businesses, and selling to corporate entities. The main difference is time! But the discussion here and growth of our customer base will be the best way to get APEX at PSDS soon.
Here was my answer to a question from Brian Lee that seems to cover a lot of what is being asked above:
Q: I'm curious about how it is able to determine what the limits are of my car at a given point in time, on a specific track with specific track conditions? I'm guessing part of it has to do with calibrating the device. But lets say while calibrating the device, I purposely take it easy e.g. driving 3/10th's of my ability perhaps due to traffic or just because its my first session of the day. Then later on, I start driving closer 7/10th's of my ability. When I start driving closer to 7/10th's I'd expect to see more green LEDs and less red. But that isn't really telling me I'm getting closer to the car's limit, correct?
A: The "calibration" you are referring to happens constantly. That's why it's considered artificial intelligence. It will constantly learn what the "limit" is based off of your driving. So if you're at 3/10 on average for a lap or so, it will think the limit is around 5/10, or whatever it has determined that "limit" to be based off of that lap, but as soon as you push it closer to 7-8/10th's, even in one corner, it has tons of new information about the limit that can be applied to every situation the car is in.
It samples over 12,000 data points/second, so any minor performance change is detected very quickly. If you're only able to drive the car at 5-6/10th's then the device may only be able to understand the limit to be 7-8/10th's. But, if you carry a lot of speed in one corner (have to enter off-line maybe, or accidentally brake too late or too little), and approach the true limit of the tire (understeer or oversteer) APEX will learn the true limit with shocking accuracy.
Through my professional coaching exploits I've noticed that most drivers, especially in cars with street tires, will approach the limit at least once during a 20-30 minute session. The limit may be found under braking, like getting into the ABS, or during acceleration (PSM/traction control intervention), but as soon as the device gets an idea of that limit, it can more accurately predict the "limit" everywhere else!
So all it takes is one "moment" and the device learns what the tire can take, and can then extrapolate that into other scenarios it's experiences around the track.
It does take at least half a lap at decent pace to build a useful simulation model for the LED display to be relevant. After 1-2 laps the feedback will be relevant to the driver. Similarly, one quick autocross run would be enough to build a robust simulation as well.
In reference to the question about tire degradation etc.. There is some "backwards learning" built into the model, but the best way to get an idea of tire deg, track temp, rain or any other grip condition that may decrease grip as the day wears on is the APEX score that is calculated in the post session data for each lap. Monitor your morning (cool track temps, fresh tires/brakes) score, vs. your afternoon score. You might find you're driving the car the same way but not quite able to achieve the same peak score as you were in cooler temps. That's showing you that the amount of grip available has decreased overall.
Joe, the reason you don't always see the LED bar full of lights is because when the LED bar is 100% full that's representing that 100% grip is available (i.e. the most/peak grip that it thinks is available that day, at that track etc..). This is likely only in a high grip scenario, a corner with good camber, or a different pavement gradient than elsewhere on the lap that provides more grip, or somewhere that your car really excels. If the device is on Beginner or Intermediate Skill Level the resolution is lower (meaning it's showing 0-100% on beginner and 75%-100% on intermediate) so you may see the LED bar full more often. If we don't see the full LED bar illuminated it means that the peak grip level cannot be achieved because either you're on an off-camber corner, on a rise or peak of some sort (braking into the corkscrew at Laguna for example), or on a low-grip surface (turns 3-4 at Laguna) etc..
@randr: Do most people watch their G-meter as they go around the track at speed? I know I don't and I've driven plenty of cars that have them. APEX is different in a few ways:
1) Calculates a lot more than just G's. It's building a friction circle model of your vehicle as you drive, and uses a 9 axis IMU, 10hz gps, accelerometer and gyro to do so. It also will interpolate the limit. The more driving you do around the 7-8/10ths mark, the closer it can get to predicting what 10/10ths actually is. It doesn't have to see the actual limit to interpolate what that limit is.
2) The display is designed to be used similarly to shift lights. Once you adapt to it being there it becomes an easily understood gauge, not one that requires your direct vision to interpret. Although a G-meter may be able to be understood by some, our goal is for the APEX display to intuitive to most, and I think the chasing LED display provides that.
@Manifold, although it is building a model of the car's potential it can only learn from the performance that you show it. So If you never get anywhere near the limit, then it is essentially showing you your limit. Does that make sense?
Thanks for the questions folks, keep them coming!
Andrew
#22
Rennlist Member
I hope you don't mind the questions but I am intrigued and I'm trying to make sure I understand what the device is showing.
So I looking at the video
and specifically at timestamp 2:50 through around 2:54, the front straight at Barber. What I see is the car gets a pretty good run out of the last corner...maybe only one red LED showing as it clips the exit curb.
Having watched the technical tutorial videos now, I get that the number of LEDs is less than "all of them" because the car is likely power limited relative to the grip potential of the tires on the front straight when accelerating.
What seemed counter intuitive to me though is that the greens steadily decrease along the straight while the number of LEDs lit in total stays the same...just more go red as the car progresses towards the braking point. I've got to imagine, and certainly the soundtrack indicates, that the driver is hard on the throttle the whole way.
The only way I could square this in my head was to imagine that the faster the car goes and the higher up the gears then the more power limited it is in terms of approaching using the total available grip for acceleration. Said another way, the horizontal power limit line on the G ellipse in Austin Gurley's whiteboard chart actually drops lower as the car goes faster in higher gears.
So if I had more power, I could light more greens but since I don't, there's really nothing the driver can do in this case to exploit the extra grip potential. Is that in fact what the device is showing or am I dangerously close to making -- ahem -- stuff up??
So I looking at the video
Having watched the technical tutorial videos now, I get that the number of LEDs is less than "all of them" because the car is likely power limited relative to the grip potential of the tires on the front straight when accelerating.
What seemed counter intuitive to me though is that the greens steadily decrease along the straight while the number of LEDs lit in total stays the same...just more go red as the car progresses towards the braking point. I've got to imagine, and certainly the soundtrack indicates, that the driver is hard on the throttle the whole way.
The only way I could square this in my head was to imagine that the faster the car goes and the higher up the gears then the more power limited it is in terms of approaching using the total available grip for acceleration. Said another way, the horizontal power limit line on the G ellipse in Austin Gurley's whiteboard chart actually drops lower as the car goes faster in higher gears.
So if I had more power, I could light more greens but since I don't, there's really nothing the driver can do in this case to exploit the extra grip potential. Is that in fact what the device is showing or am I dangerously close to making -- ahem -- stuff up??
#23
You're on it boxer-11, that's exactly why! Also, the device does model aero as well. Drag in particular, so as the car creates more drag it's going to slow at a faster rate. Basically, the car is accelerating slower than the device knows it can (when it's in a lower gear, at lower speed).
If you had unlimited horsepower, it would be possible to overlap the green and the red in a straight-line.
Would it be more helpful if the display showed how close you were to your previous best acceleration on the straights instead of ultimate grip?
Great question!
If you had unlimited horsepower, it would be possible to overlap the green and the red in a straight-line.
Would it be more helpful if the display showed how close you were to your previous best acceleration on the straights instead of ultimate grip?
Great question!
#24
Rennlist Member
Awesome...that's a lot of information in a very small space...just need to grok all the nuances of what it's showing you.
I'd say yes.
In fact... At the risk of cluttering the display that makes me wonder about a third color (which I think you already have the hardware for if I recall from the video that covered start up). I would imagine that once you have some data in the bag you could show previous best for all regions of the track. Say:
Red (for potential) overlapped by orange (or blue or something...to show previous best grip exploited at this point) overlapped by green (to show current exploit). So just add the middle color "layer" to show where I've been doing better previously... In other words, on my best performance at this point on the track I'd see green and maybe some red if there's more potential grip; while if I'm doing less well this lap at this location I'd see green, some orange and maybe red if there's potential I haven't ever used at this spot. [afterthought: hmm, this might need options like show me either best ever at this spot or show me this spot on my best lap ever or so far in this session...again correlating to what predictive timers do perhaps; best ever at the precise spot might be beyond reach in the same way as theoretical best lap pasted together from short sector times].
If you did something like that you'd almost have something in the spirit of predictive lap timers only set up to tell me better/worse grip exploit at the given point on the track.
In fact... At the risk of cluttering the display that makes me wonder about a third color (which I think you already have the hardware for if I recall from the video that covered start up). I would imagine that once you have some data in the bag you could show previous best for all regions of the track. Say:
Red (for potential) overlapped by orange (or blue or something...to show previous best grip exploited at this point) overlapped by green (to show current exploit). So just add the middle color "layer" to show where I've been doing better previously... In other words, on my best performance at this point on the track I'd see green and maybe some red if there's more potential grip; while if I'm doing less well this lap at this location I'd see green, some orange and maybe red if there's potential I haven't ever used at this spot. [afterthought: hmm, this might need options like show me either best ever at this spot or show me this spot on my best lap ever or so far in this session...again correlating to what predictive timers do perhaps; best ever at the precise spot might be beyond reach in the same way as theoretical best lap pasted together from short sector times].
If you did something like that you'd almost have something in the spirit of predictive lap timers only set up to tell me better/worse grip exploit at the given point on the track.
Last edited by boxer-11; 08-08-2017 at 09:33 PM. Reason: afterthought...
#25
I can see how this would work and learn the tire limits quickly (see friction circle below) - essentially if you "noob" brake (e.g. hard one off to ABS - not firm to trail) from a fast straight into a sharp corner you will pretty much get to the limit of the tires.
The example below is a series of friction circles (1) fresh P Zero N1 Vs (2) worn P Zero N1
The bottom of the Y-axis under hard braking shows the tire limit to be 1.45 g, which more or less matches the peak cornering g which is achieved close to the apex of a corner and where max yaw is achieved.
I like the principle and can see its value - but also note that the friction circle of the worn tires is terrible relative to the fresh tires.
The question I have is does the system allow for the degradation of the tires within a series of laps e.g. as the diameter of the friction circle decreases as the tires wear/over cook?
The example below is a series of friction circles (1) fresh P Zero N1 Vs (2) worn P Zero N1
The bottom of the Y-axis under hard braking shows the tire limit to be 1.45 g, which more or less matches the peak cornering g which is achieved close to the apex of a corner and where max yaw is achieved.
I like the principle and can see its value - but also note that the friction circle of the worn tires is terrible relative to the fresh tires.
The question I have is does the system allow for the degradation of the tires within a series of laps e.g. as the diameter of the friction circle decreases as the tires wear/over cook?
#26
randr, APEX does account for tire deg/ grip level degradation. It's built in the model to understand progressive degradation. If however there is a major and sudden grip change (lots of rain, for instance), the best way to get accurate information from APEX is to reset it (turn on and off).
What makes APEX different is not the data we collect it's the sample rate, the resolution, and the filtering that we do with the data. We're not showing you G-Sum, the light bar represents a % of how much of the car's potential grip you're using! We're not providing a tool for analyzing numbers, it's a tool to help make you faster.
On a side note I wish I was on the GT3 owners forum because I owned a GT3. I'm assuming my 83' 944 doesn't quite count... haha
What makes APEX different is not the data we collect it's the sample rate, the resolution, and the filtering that we do with the data. We're not showing you G-Sum, the light bar represents a % of how much of the car's potential grip you're using! We're not providing a tool for analyzing numbers, it's a tool to help make you faster.
On a side note I wish I was on the GT3 owners forum because I owned a GT3. I'm assuming my 83' 944 doesn't quite count... haha
#27
Rennlist Member
Not being a big data guy I find this intriguing. Sorry if I may have missed this, but is there a recommendation for a good video camera that would work well with this device for review after the session? The camera doesn't have to have all the data collection features (like Aim) just a good easy to use camera.
Thanks
Thanks
#28
Rennlist Member
It's a great idea and I can totally see how it can be helpful.
That said the G-loads afforded in different corners often vary widely even in the same lap. Peak GSUM depends on things like road camber, tarmac, aero etc. etc. Also calibrating the right baseline for each session will be difficult as tires and surface change...
Could be a good rough guide to train up one's internal butt feel I suppose?
That said the G-loads afforded in different corners often vary widely even in the same lap. Peak GSUM depends on things like road camber, tarmac, aero etc. etc. Also calibrating the right baseline for each session will be difficult as tires and surface change...
Could be a good rough guide to train up one's internal butt feel I suppose?
#29
Check out this link futurz, this was shot with a GoPro Hero 4:
It's all about playing with the frame rate so the LED's don't appear to flutter. But most in-car systems capture the lights well if the right frame rate is used. I believe this is 60 FPS or better in this video. In my experience the higher the frame rate the better for capturing the LED's.
CRex, absolutely! "That said the G-loads afforded in different corners often vary widely even in the same lap. Peak GSUM depends on things like road camber, tarmac, aero etc. etc. Also calibrating the right baseline for each session will be difficult as tires and surface change..."
That's what APEX technology does, it takes over 12,000 measurements/second with a 9 axis IMU, 10hz GPS, Accelerometer, and Gyro. Then uses that information to build a simulation of the vehicle's capability. The secret sauce is not taking measurements, it's what we do with those measurements that make the technology highly accurate and adaptable!
It's all about playing with the frame rate so the LED's don't appear to flutter. But most in-car systems capture the lights well if the right frame rate is used. I believe this is 60 FPS or better in this video. In my experience the higher the frame rate the better for capturing the LED's.
CRex, absolutely! "That said the G-loads afforded in different corners often vary widely even in the same lap. Peak GSUM depends on things like road camber, tarmac, aero etc. etc. Also calibrating the right baseline for each session will be difficult as tires and surface change..."
That's what APEX technology does, it takes over 12,000 measurements/second with a 9 axis IMU, 10hz GPS, Accelerometer, and Gyro. Then uses that information to build a simulation of the vehicle's capability. The secret sauce is not taking measurements, it's what we do with those measurements that make the technology highly accurate and adaptable!
#30
Check out this link futurz, this was shot with a GoPro Hero 4: https://www.youtube.com/watch?v=PU6A6Ci8AkU
It's all about playing with the frame rate so the LED's don't appear to flutter. But most in-car systems capture the lights well if the right frame rate is used. I believe this is 60 FPS or better in this video. In my experience the higher the frame rate the better for capturing the LED's.
CRex, absolutely! "That said the G-loads afforded in different corners often vary widely even in the same lap. Peak GSUM depends on things like road camber, tarmac, aero etc. etc. Also calibrating the right baseline for each session will be difficult as tires and surface change..."
That's what APEX technology does, it takes over 12,000 measurements/second with a 9 axis IMU, 10hz GPS, Accelerometer, and Gyro. Then uses that information to build a simulation of the vehicle's capability. The secret sauce is not taking measurements, it's what we do with those measurements that make the technology highly accurate and adaptable!
It's all about playing with the frame rate so the LED's don't appear to flutter. But most in-car systems capture the lights well if the right frame rate is used. I believe this is 60 FPS or better in this video. In my experience the higher the frame rate the better for capturing the LED's.
CRex, absolutely! "That said the G-loads afforded in different corners often vary widely even in the same lap. Peak GSUM depends on things like road camber, tarmac, aero etc. etc. Also calibrating the right baseline for each session will be difficult as tires and surface change..."
That's what APEX technology does, it takes over 12,000 measurements/second with a 9 axis IMU, 10hz GPS, Accelerometer, and Gyro. Then uses that information to build a simulation of the vehicle's capability. The secret sauce is not taking measurements, it's what we do with those measurements that make the technology highly accurate and adaptable!
Just kidding with you guys... cool product concept.