41 replies to this topic

[Steve Scheifler]

If you get Ross Bentley’s emails you probably already know that Garmin just released their new AI driving analyzer and coach, the Catalyst. Best that you read the marketing materials rather than me try to hit all the features but they are not (yet) competing directly with a full data system and advanced PC analysis software. This is a stand-alone tablet plus camera that records your laps then applies AI to identify “opportunities” to lower your lap times including real-time audio coaching for some things. If it were by some unknown company I would be thinking just another gimmicky gadget, but by multiple accounts Garmin has been working on this a long time. Anyway, it’s very focused, no additional inputs or add-ons but there’s still quite a lot to it. Even if all you want is a good predictive lap display and video with basic overlay, this isn’t a bad deal. At $1000 it’s the same as an AiM SmartyCam, JUST the camera.

https://discover.gar...en-US/catalyst/

[Alberto]

Seems to offer similar functionality as the AIM Solo but a nicer display?  Is there some sort of software package for analysis like AIM offers?

[Steve Scheifler]

It is not intended for PC based analysis and is both more and less than the Solo. It boils even the basic data down to the absolute basics, auto-comparing segments from your session of laps looking for what worked best then formulating specific recommendations for what to do with regard to speed, braking and line, what they term Opportunities. Then you can look at graphical representations of the recommendations and see both in graphs and synched images and video the difference between one lap and another, all compiled for you rather than trying to interpret the data yourself.

I think one of the best experiments would be to run it in parallel with a Solo then look at the AiM data to see if roughly the same conclusions are reached. Surely they did a lot of that during development of the AI.

One of the things unique to this, at least as I read it, is that the AI includes the video “view” as data input when evaluating your position on track etc. This is surely a first in consumer products of this type and if it’s legit then quite interesting.

Delivery of mine got pushed to end of the day but I’ll be posting what I learn with it on the Catalyst FB group over the next week.

[manthony121]

There's not much technical info available that I could find, but the Catalyst sounds "interesting".  One inherent limitation with GPS data acquisition is that it can only be accurate to with a few feet.  That is plenty good enough for lap timing, calculating g-forces, etc, but for refining your driving line, it is not precise enough.  The Catalyst, near as I can tell from the marketing materials, seems to also use a high resolution video camera to determine your driving line with improved precision compared to GPS alone.  I suppose this is a spin off of all the work going on to use video to control self-driving cars, lane departure warning systems, etc.

[gerglmuff2]

There's not much technical info available that I could find, but the Catalyst sounds "interesting".  One inherent limitation with GPS data acquisition is that it can only be accurate to with a few feet.  That is plenty good enough for lap timing, calculating g-forces, etc, but for refining your driving line, it is not precise enough.  The Catalyst, near as I can tell from the marketing materials, seems to also use a high resolution video camera to determine your driving line with improved precision compared to GPS alone.  I suppose this is a spin off of all the work going on to use video to control self-driving cars, lane departure warning systems, etc.

autocrossers have been using 10hz gps to compare lines on an autocross course for 10 years. its accurate enough to compare line choice on a race track. 

[Steve Scheifler]

Well there’s accuracy in an absolute sense and there’s repeatability. This purports to be at the upper end of both for systems not using ground-based correction so under ideal conditions (the flat plains of Kansas) it may have sub-meter repeatability during the course of a session at least. But I think your surmise is probably correct.

[Steve Scheifler]

First track test:

We had one in a car for the PDE at Gateway this afternoon but the driver managed only two interrupted sessions before someone in the intermediate group moved a wall with their Honda and ended the day. So total laps was about 17 of which only a few were at any pace. As the driver was also an instructor for other sessions he had little opportunity for me to demonstrate the capabilities so we didn’t get much review done. After session two I had accepted the coach voice prompts for the three identified “opportunities” to be active for the next session but that never happened. Oh well, we did at least get through the basic install & setup, recoded laps in two sessions and got “opportunity” feedback of several types for braking and line. The driver may choose to chime in here on his initial observations and experience though it was very limited.

Next up will be a very different test, twin 10 hour enduros this coming weekend. There really isn’t time to do this right so it might turn into an exercise in frustration, but there are so few opportunities left this season that we don’t want to pass one up. The most experienced driver is running first so the temptation is to let him put down a series of clean laps at the target pace then leave it running as a single session for all the others to work from. But of course that means no review during the race. So I’m sending a second unit to give them the option to swap it every driver change, allowing review of the most recent laps by the driver immediately after their session. First time out with it I favor the swap approach at least the first day, then decide which way to go on Sunday. I won’t be there myself but I’m hoping that their car is trouble free and circumstances allow them to work with the Catalyst.

[mdavis]

This device looks pretty slick.  I'm curious as to how you mounted it.  I'm not a huge fan of the suction cup method.  Looks like there is a solid mounting method but still requires multiple arm "extensions" so mounting it flat somewhere does not seem to be an option.  Also the camera isn't moveable either with a single adhesive mount- not sure why they did that.  I'm sure there is a work around- thoughts on that one?

[Steve Scheifler]

Because I’m moving it around a lot I’m using the suction cup. I agree that it is an iffy method particularly for something of any size and weight but done correctly the one they provide is very strong and I doubt it would come off short of fracturing the glass. A small screw/bolt down base comes with it and I would absolutely use that instead for a more permanent install but that won’t happen in a lot of track days cars.

My bigger issue with the mount system is that the tablet lacks any means of tethering it. It would have been a simple matter to incorporate a small tether slot in each end and should have been a no-brainer at the moment they chose a magnetic connection from tablet to base.

[mdavis]

What did you do with the camera for mounting- assuming you are using it and didn't just use the adhesive?

[Steve Scheifler]

What did you do with the camera for mounting- assuming you are using it and didn't just use the adhesive?

The camera is not optional. I used the provided tiny adhesive mounts for now. There are small suction cup versions out there that probably use the same ball size but for now it’s simplest to just stick it on the windshield with the provided mounts. Easy enough to replace the adhesive and use them again, the camera is small and very light.

[mdavis]

Any more information on this from the enduros?

[Steve Scheifler]

No. That didn’t go well. During the first driver change the camera was bumped off alignment by a lot. The AI is built around the camera “view” so nothing useful was gained after that other than lap times and the continuous +\- delta to best lap display.

I’ve lent them out several times since but in almost every case the users had issues with their car or other circumstances limited the testing. I still like the idea of having them as loaners but despite being fundamentally simple compared to traditional systems it requires a little time and attention to detail at least initially. I haven’t been able to be there myself to assist. One of them was at MidO for the IT Spectacular weekend but the car had a diff failure on the first lap of practice. That was fixed and it went to Hallett for a Comma weekend. There he finally got good data and found the post-session reviews helpful but never turned on the live advanced voice coaching, which should be it’s big strength and differentiator. So, next season I guess.

Meanwhile the FB group is getting better feedback with a number of average Joe (unbiased) users very impressed and claiming rapid improvements using the advanced voice coach.

There’s still a lot of misunderstanding of what it is, and isn’t, but most of those who “get it” seem pleased with their purchase.

[38bfast]

Is it measuring to best in session or can you load best from other session to compare to? Can you have your Shoe driver do some laps and then when the no so shoe get in he get the reference back to the fast driver? 

[Steve Scheifler]

Is it measuring to best in session or can you load best from other session to compare to? Can you have your Shoe driver do some laps and then when the no so shoe get in he get the reference back to the fast driver?

It is organized by car & driver. So for a given car-driver you can do some comparisons between two sessions in review mode. That option is not readily apparent and is definitely not the focus, at least in their minds, but it is there.

Throughout a weekend (stated as being ~ 4 days) at a given track it is using data from prior sessions (divided into wet vs dry assuming you define that before each session) for learning the track, car and driver. So yes, if you put a faster driver in the car using your profile for a session to teach it what the car can do, it will then use that information to encourage you to match that performance, believing you had already achieved it but inconsistently. Not surprisingly they publicly warn against such tactics.

[bmarshall1]

I purchased a Catalyst but have yet to use it.  If things work out, I will have the opportunity at Sebring for their Turkey Trot event.  I'll post details and experience.

[38bfast]

Steve I am on the fence to how much value there is over a Solo. A lot of information going on while you're driving with the Garmin. 

[Steve Scheifler]

Steve I am on the fence to how much value there is over a Solo. A lot of information going on while you're driving with the Garmin.

Well, if a person definitely wants to learn detailed data analysis and has time to master that or someone to do it for them even between sessions, and has access to truly comparable data, then the Solo or equivalent is a great entry-level tool (though I’d argue only if accompanied by synced video which IMO is critical to weeding out the countless real-world anomalies that occur during most sessions). But from my experience most end up glorified lap timers and YouTube content generators and most of those who invest the time to use the software are likely as not to misinterpret what they see. The potential is inarguably enormous but that potential is rarely even close to fully realized.

By contrast the Catalyst attempts to do the analysis for you then turn the results into digestible recommendations with both real-time voice coaching and post-session text and illustration.
So a lot going on while you are driving, yes, but that’s optional and probably best used while testing and qualifying rather than dicing in a group during a race. So far I haven’t heard or read anyone complaining about the voice coach being an unwelcome distraction, and it seems to be far more beneficial in general than the post-session review. Both do require that the driver NOT be too consistent because it makes no assumptions about the car’s capabilities, or the driver’s for that matter. It needs different lines, brake points etc. to compare and analyze in combinations for the best overall result. It definitely has limitations, but I’ll gladly wager that almost anyone new to data can lower lap times more and faster using the Catalyst over any other device. That doesn’t make it the last word in race data or coaching by any means, or a replacement for full data systems IF they are actually utilized, but I’ve seen and read enough to say I would recommend it as the first $1,000 someone should spend in that direction.

[bmarshall1]

I had the opportunity to use mine a little last weekend.  I had set it up at home (about 10 minutes entering my info), and once to the track it was plug and play.

 

Important - I did not have the ability to hook it up to my race comms,  I am currently looking for the right combo of adaptors and extensions to allow me to use both my race radio and hear the Catalyst at the same time.  I think it should be pretty easy - (1) mono Y connector and (1) mono (6 ft?) extension cord. Likely under $20 total on Amazon.  Maybe Steve already knows what is needed and can share?

 

Anywho... I Hit the track and it starts automatically and after a lap or two starts your over/under as compared to you last laps.

 

That night I reviewed the info pretty in depth. It compares your average line to you best and give you time differences "you can save .54 seconds" etc... It shows consistency and a few other items I can't remember at this time.  It shows braking, acceleration, lines through the corners and gives you 'coaching' "Turn in sooner, apex sooner, brake later". (I could have told you that  )   But it's nice to actually see it and then look at the line they suggest and review a 'stitched' together lap to see where on the track you should be.  Unfortunately I blew a head gasket on Friday's practice and did not get to use it further.  I also was not able to use the in car coaching feature, which I think is the best feature, real time track guidance.

 

Next time I get out there I'll share a little more

[bmarshall1]

@Steve Scheifler  I tried to send you a PM but you weren't able to receive, maybe your inbox is full?