What makes a better driving coach, a human or an AI data analyst?

Anyone who’s seen any of the “Terminator” films knows that the AI rebellion that will wipe out all humans is inevitable. Heck, AI is already coming for some jobs–like conducting tedious legal research or writing nonsensical car magazine stories–but now it may be sticking its robotic nose a little too far into a tent pretty close to home.

That’s right, …

Read the rest of the story

After we finished that day, I recall discussing a third option: Both.

A human driving instructor fits the car of a student with a Garmin Catalyst (or similar), sends the student out and then interprets the recorded data with the student to improve lap times.

This way, the instructor could potentially instruct multiple students–and without having to ride along in any cars.

A+ Futurama references. 

Personally I don't do that well with classroom coaching or data. I seem to learn best from chasing another similar car that's slightly faster than me. Not only do you get an exact feel for their line and braking points, but you know that if they can do it then you can do it too. It's also super obvious if there are spots where you're gaining on them. This type of lead follow is easier to do on iRacing than IRL but it is a great help in both areas. That said, it's best for finding tenths, not 16 seconds!

Colin Wood said:

After we finished that day, I recall discussing a third option: Both.

A human driving instructor fits the car of a student with a Garmin Catalyst (or similar), sends the student out and then interprets the recorded data with the student to improve lap times.

This way, the instructor could potentially instruct multiple students–and without having to ride along in any cars.

a few years ago I had an instructor who was willing to take my car out for a session with me in the passanger seat.  After this I was able to compare the two laps (his fastest vs my fastest).  I did that /w a app based data collection and the only way that I could compare them was to watch the video side by side, with a catalyst running it would be very possible for me to drive and then compare my fastest lap to an instructors fastest lap.  If you do it on the same day /w the same car in differing sessions you should be able to determine line differences, braking points, etc.  

the other benefit of an instructor taking your car for a drive is that its possible to see the potential of the car which is ALWAYS more than I am extracting out of it and enables me to avoid spending money on parts and instead focus on track time.

After we did this story I got the chance to have J.G. right seat with me at the FIRM in our project MK7 GTI and I think both the Catalyst and a real coach is the answer. Like in the story the Catalyst helped a lot to get me going quicker but having J.G. there to explain why I was going quicker was super helpful and it was nice to be able to talk the track through with someone vs just getting instructions from a computer. 

Chris Tropea said:

After we did this story I got the chance to have J.G. right seat with me at the FIRM in our project MK7 GTI and I think both the Catalyst and a real coach is the answer. Like in the story the Catalyst helped a lot to get me going quicker but having J.G. there to explain why I was going quicker was super helpful and it was nice to be able to talk the track through with someone vs just getting instructions from a computer. 

i think this is also key, having a good instructor.  a good teacher has to meet the student where they are in the learning journey.  if you're 5-6 seconds off pace the advice needed is probably different than if you're .5 or .6 seconds off pace.  

I'm just here for the Futurama references.

A few of the folks I TT with have a Catalyst. I've heard some say the AI coach is probably more useful for beginning drivers but a couple of very experienced (and fast) drivers have said that the AI has helped them on new tracks as well as ones they've run a lot of laps on.

It will be interesting to see how AI evolves for applications like this. It's in its infancy now, who knows what the future will bring.

FarmerTed said:

I'm just here for the Futurama references.

A few of the folks I TT with have a Catalyst. I've heard some say the AI coach is probably more useful for beginning drivers but a couple of very experienced (and fast) drivers have said that the AI has helped them on new tracks as well as ones they've run a lot of laps on.

It will be interesting to see how AI evolves for applications like this. It's in its infancy now, who knows what the future will bring.

Yeah I definitely agree that this technology has some real potential in this space. This particular application does a good job at leveraging what I think is one of the strengths of  the current state of AI, which is data aggregation and comparison.

I still think it's a more useful tool for drivers of some level of experience, though. A totally green novice is going to be asked to do things that they don't fully grasp how to attempt, or even what some of those attempts entail. So I think for the true noobs, a meat instructor is better able to assess not only their skill level but their knowledge level and knowledge of how to execute what skill set they have. 

JG Pasterjak said:

A meat instructor is better able to assess not only their skill level but their knowledge level and knowledge of how to execute what skill set they have. 

I think we need to make you a t-shirt that says "meat instructor" on it for the next time you're in a driving coach role.

My son's first track experience was during COViD, so they were doing lead-follow.  He was super green, and didn't really understand the goals/intent of the instructor, so the lead-follow with signals, etc., was sub-optimal, to say the least.  I think this is like turning a total novice loose with Garmin.  As one of the first commenters said, a combination of good data and a ready analysis tool, COMBINED with an in-car instructor, is the key to getting a novice off the ground and flying solo competently.  It seems like the Garmin is more useful for at least a moderately experienced driver who at least understands what they should be doing with the suggestions and guidance they're getting from the Garmin.

HiFiGuy said:

My son's first track experience was during COViD, so they were doing lead-follow.  He was super green, and didn't really understand the goals/intent of the instructor, so the lead-follow with signals, etc., was sub-optimal, to say the least.  I think this is like turning a total novice loose with Garmin.  As one of the first commenters said, a combination of good data and a ready analysis tool, COMBINED with an in-car instructor, is the key to getting a novice off the ground and flying solo competently.  It seems like the Garmin is more useful for at least a moderately experienced driver who at least understands what they should be doing with the suggestions and guidance they're getting from the Garmin.

my first track day was the year before covid and my second was during covid, thank you; I now have a new racer excuse... "nobody taught me..."

I think they compliment each other. The Garmin alone can only help you refine what you're already doing. If you're way off line it can only help so much. An instructor can help get someone on the right line a lot quicker because they know where it is and can actually tell you. An instructor otoh will get you up and running, but without data its hard to be objective about if what you're doing is working. Good enough when you are super far off the pace, but difficult when you're trying to find tenths here and there. Together they give an instructor the data they need to make good suggestions further into the learning process. It allows you to still have feedback even if an instructor isn't available all the time. Its easier to spend a few minutes going over someone's session than being in car with them all the time. Heck with the app features you can share after the fact and remotely now.

Where it really falls down is the depth of the data logging. No basic obd2 data. No way to export the telemetry to external programs for deeper analysis.  Other minor quibbles too. The camera quality is horrible for sharing online. No support for autocross even as a basic timer. The speaker is entirely too quiet to be heard on track with the windows down. All the telemetry and sharing options in the companion app really should be available on the device itself.

I'm starting to get to the point where I want some of those more advanced features or I need to move on to a device that's a little less friendly in order to get it.

In reply to Colin Wood :

Agreed, but you don't need a $1,000 Garmin Catalyst for that. There are any number of good GPS data acquisition devices (including your Phone, which has limitations) that cost significantly less and offer way more flexibility in post-session analysis than the Garmin does.

A good coach is someone who can understand how the student learns, is able to identify areas for improvement and is able to communicate efficiently while being able to interpret data and pinpoint those areas for improvement. The point in the article about the coach being able to address the how and why, and not just the what, is spot on!

Data --> Insight --> Knowledge --> Wisdom

A data/video collection tool gets you in the game at Stage 1.  Self-coaching and digital analysis tools can get you to Stage 2, and maybe into Stage 3 if you're really determined.  A human coach can get you the rest of it, because IMO it doesn't happen without somebody to share first-hand experience.

I used to do track days at Waterford Hills with the MGB GT. Was a lot of fun but I was pretty slow and plateaued . Was not sure if it was me or the car. So I went out with an instructor who was also a very fast driver. After a couple of laps he told me that  there was nothing technically wrong with my driving, but I would never be much quicker because my imagination was too good. "You are focused on what could go wrong  at any moment, rather than how to set up for the next corner" "That will forever keep you from using that last inch of track surface needed to go fast". 

He was right. "Never go on a track unless you are mentally and emotionally prepared to leave your car and your health in the dumpster on the way out" No AI could have detected what he did.

NOHOME said:

I used to do track days at Waterford Hills with the MGB GT. Was a lot of fun but I was pretty slow and plateaued . Was not sure if it was me or the car. So I went out with an instructor who was also a very fast driver. After a couple of laps he told me that  there was nothing technically wrong with my driving, but I would never be much quicker because my imagination was too good. "You are focused on what could go wrong  at any moment, rather than how to set up for the next corner" "That will forever keep you from using that last inch of track surface needed to go fast". 

 

He was right. "Never go on a track unless you are mentally and emotionally prepared to leave your car and your health in the dumpster on the way out" No AI could have detected what he did.

counter point, a track day isn't a race...  I would rather have fun while managing my risk, there is no winning at a track day, only losing.