Garmin Catalyst Driving Performance Optimizer

[jfritz27]

This may have an obvious answer, but how do these systems "know" what time may be left on the table? Specifically, there seems to be the "unknown" variable (based on your tires, suspension setup, etc) of whether the car is going to hold if I carry X amount more speed through this or that corner by braking later or doing whatever the coaching tells me to do. Do these packages somehow use fancy algorithms of being able to "predict" what your car is capable of by analyzing other corners (max G established in some other corner, for example) or some other technique (or input parameters)?

Just curious

[paffinity]

Originally Posted by jfritz27

This may have an obvious answer, but how do these systems "know" what time may be left on the table? Specifically, there seems to be the "unknown" variable (based on your tires, suspension setup, etc) of whether the car is going to hold if I carry X amount more speed through this or that corner by braking later or doing whatever the coaching tells me to do. Do these packages somehow use fancy algorithms of being able to "predict" what your car is capable of by analyzing other corners (max G established in some other corner, for example) or some other technique (or input parameters)?

Just curious

From my experience it's just based on previous runs, whatever your best segment time was it stitches together and says that's your potential time. If you're always taking the wrong line it's not going to help you. So I don't think this is for someone new to DE but rather someone that is willing to try out different lines so that there's enough information for the system to learn.

[jfritz27]

Thanks.... it is interesting to consider then exactly how useful these things are. Not saying they're not, but by that logic you basically have to set a faster time for the thing to be able to know that you can set a faster time (!?). I understand the concept of putting together an ideal best lap from individual best sectors, so I guess that's where the value is. But at the end of the day, if you want to push your times on any one sector, it sounds like the fundamental limitations of the method can't really help you (which is understandable). Unless I'm missing something.

Just trying to get a sense of how useful this would be.

[MagnusT]

It worked better at Mantorp than Sviestad. Maybe it had a hard time understanding some corners there? I don't know.

Garmin Catalyst - Trackday #2 Mantorp (FULL unedited session 1)

Filming the Catalyst using Racechrono for reference. You can see what the Catalyst shows on it's screen, what the Catalyst says and also see some extra data recorded with Racechrono for reference/comparison to the Catalyst and the driving.

Three warmup laps. Really cool cars on the track! Formula open wheelers, GT40, Ariel and some Porsche GT cars. I start pushing from lap 4. Still some time on the table when I finish the session to what I can do normally. More in session 2!

You be the judge, do you think the Catalyst is giving me good coaching tips this trackday?

[fanatic1]

Originally Posted by MagnusT

It worked better at Mantorp than Sviestad. Maybe it had a hard time understanding some corners there? I don't know.

Garmin Catalyst - Trackday #2 Mantorp (FULL unedited session 1)

Filming the Catalyst using Racechrono for reference. You can see what the Catalyst shows on it's screen, what the Catalyst says and also see some extra data recorded with Racechrono for reference/comparison to the Catalyst and the driving.

Three warmup laps. Really cool cars on the track! Formula open wheelers, GT40, Ariel and some Porsche GT cars. I start pushing from lap 4. Still some time on the table when I finish the session to what I can do normally. More in session 2!

You be the judge, do you think the Catalyst is giving me good coaching tips this trackday?

Awesome tire temperature monitors!

[Gomeler]

Uhh.. yeah. I would like to know more about where it was getting that tire temperature data. I presume an IR sensor in the wheel well?

[MagnusT]

Originally Posted by Gomeler

Uhh.. yeah. I would like to know more about where it was getting that tire temperature data. I presume an IR sensor in the wheel well?

DIY kit I designed a while back. $100 per wheel DIY.

https://github.com/MagnusThome/RejsaRubberTrac

[MagnusT]

Here's the main video from all my testing so far.

Comment if you think it is giving relevant advice or not!

[DRLane]

Originally Posted by MagnusT

DIY kit I designed a while back. $100 per wheel DIY.

https://github.com/MagnusThome/RejsaRubberTrac

Thank you!!!

[paffinity]

Originally Posted by MagnusT

Here's the main video from all my testing so far.

Comment if you think it is giving relevant advice or not!

Awesome, it does seem like it generally helps. I'm sure with less traffic you would have matched your PB. I did notice the weird lines it suggest at times, so that may be a bit confusing to others.

[nicknaz]

Originally Posted by MagnusT

Here's the main video from all my testing so far.

Comment if you think it is giving relevant advice or not!

Thanks for sharing Magnus.

Personally, I don’t think it is relevant, and I think folks should save up and get AiM with sensors plus smartycam, get someone faster to do hot lap to put data and the aim, get coaching on how to use aim.

Great idea though, and thank you so much for sharing and beta testing for everyone

[MagnusT]

Originally Posted by nicknaz

Thanks for sharing Magnus.

Personally, I don’t think it is relevant, and I think folks should save up and get AiM with sensors plus smartycam, get someone faster to do hot lap to put data and the aim, get coaching on how to use aim.

Great idea though, and thank you so much for sharing and beta testing for everyone

They are for sure following two very different approaches.

[Gomeler]

Originally Posted by MagnusT

DIY kit I designed a while back. $100 per wheel DIY.

https://github.com/MagnusThome/RejsaRubberTrac

*looks at his pile of arduino parts* you've done what I quickly forgot about. Going to assemble one of these and try it out.

[MagnusT]

Originally Posted by Gomeler

*looks at his pile of arduino parts* you've done what I quickly forgot about. Going to assemble one of these and try it out.

When Covid hit it seems all IR sensor manufacturers focused on sensors for common handheld thermometers and other sensor types were put on hold. So they became out of stock in many places and the ones available went up in price. But it is a couple of months since I checked last, so I don't know availability right now.

[dvq]

Originally Posted by jfritz27

This may have an obvious answer, but how do these systems "know" what time may be left on the table? Specifically, there seems to be the "unknown" variable (based on your tires, suspension setup, etc) of whether the car is going to hold if I carry X amount more speed through this or that corner by braking later or doing whatever the coaching tells me to do. Do these packages somehow use fancy algorithms of being able to "predict" what your car is capable of by analyzing other corners (max G established in some other corner, for example) or some other technique (or input parameters)?

Just curious

From my understanding this and many other systems that have accelerometers simply store samples that represent an internal G-G diagram. If you think about your combination of longitudinal acceleration and lateral acceleration you can place a point on a graph which represents your car's total acceleration vector.

If you then record all of these samples and store the maximum values of that total acceleration then you now have a series of maximum acceleration vectors that you know your car is capable of. When you start sliding those acceleration sums naturally drop.

That's why this system, apex pro, and many others that have focuses on letting you know how much available grip you have take 2-3 laps to warm your tires up and figure out how much grip you have available.

Anyone who has more knowledge on the subject feel free to correct me if anything I said was wrong, I'm in the process of figuring this out myself. It seems to make sense to me and this photo from the AIM manuals helped me understand that G-G diagram.

The rest of the calculations from a timing segment point of view are based on what your bests are -- the combination of G-G diagram, line, and GPS come together to form their secret sauce which tells you the timing you could improve on.

[MagnusT]

With a good 9-dof accelero- and position-sensor you can quite surely also detect both understeer and oversteer as an addition to the general traction circle

[jfritz27]

Originally Posted by dvq

From my understanding this and many other systems that have accelerometers simply store samples that represent an internal G-G diagram. If you think about your combination of longitudinal acceleration and lateral acceleration you can place a point on a graph which represents your car's total acceleration vector.

If you then record all of these samples and store the maximum values of that total acceleration then you now have a series of maximum acceleration vectors that you know your car is capable of. When you start sliding those acceleration sums naturally drop.

That's why this system, apex pro, and many others that have focuses on letting you know how much available grip you have take 2-3 laps to warm your tires up and figure out how much grip you have available.

Anyone who has more knowledge on the subject feel free to correct me if anything I said was wrong, I'm in the process of figuring this out myself. It seems to make sense to me and this photo from the AIM manuals helped me understand that G-G diagram.

The rest of the calculations from a timing segment point of view are based on what your bests are -- the combination of G-G diagram, line, and GPS come together to form their secret sauce which tells you the timing you could improve on.

Interesting, thanks. But again, it would seem it can only work with those maximum acceleration vectors that it was able to record. If you were to intentionally not push, then presumably it wouldn't be able to predict that you could push harder. That's obviously inherently intuitive on some level I suppose. I guess maybe it has some iterative process whereby it will say something like, "let's try a 0.5% increase in X, Y, or Z" and see how the car reacts, then assuming no shenanigans it rinses and repeats.

[Suds]

Yeah, my experience is it uses your best times to suggest quicker driving lines.

Originally Posted by jfritz27

Interesting, thanks. But again, it would seem it can only work with those maximum acceleration vectors that it was able to record. If you were to intentionally not push, then presumably it wouldn't be able to predict that you could push harder. That's obviously inherently intuitive on some level I suppose. I guess maybe it has some iterative process whereby it will say something like, "let's try a 0.5% increase in X, Y, or Z" and see how the car reacts, then assuming no shenanigans it rinses and repeats.

[Gomeler]

Originally Posted by MagnusT

When Covid hit it seems all IR sensor manufacturers focused on sensors for common handheld thermometers and other sensor types were put on hold. So they became out of stock in many places and the ones available went up in price. But it is a couple of months since I checked last, so I don't know availability right now.

You're not kidding. Nothing by Melexis is in stock anywhere, with limited/no availability from the factory. Maybe supply will open up in 2021

[Gomeler]

Originally Posted by Gomeler

You're not kidding. Nothing by Melexis is in stock anywhere, with limited/no availability from the factory. Maybe supply will open up in 2021

Found a few thermopile arrays that might be a suitable alternate, albeit considerably lower resolution(8x8 vs 16x2). But might be workable for now, until the higher resolution sensors are restocked. Going to fiddle with this a bit, seems like a fun project.

[MagnusT]

Information directly from Adam Spence the project lead at Garmin for the Catalyst

https://speedsecrets.com/160-adam-sp...faster-driver/

[marvinp]

Originally Posted by MagnusT

Here's the main video from all my testing so far.

Comment if you think it is giving relevant advice or not!

Pretty cool, thanks for sharing! What does it mean when it says "keep pushing." It said it during the final turn. Does it mean "get on the gas"?