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

Wanna Kill All Humans?

No, I don’t hate robots. Then again, none have hunted me or my loved ones for sport, so it suffices to say we have a mutually respectful arrangement for the time being. I do occasionally help folks go faster, though, whether it’s their first time on track or they’re seeking more confidence and accompanying speed. 

I’ve also been known to use artificial intelligence for these exercises. But could that tech replace someone like me? The question intrigued me: If someone was looking to get on track for the first time, could robot claws outperform human hands?

The exercise we proposed was simple: Let’s take two drivers with little track experience to our official test course, the Florida International Rally & Motorsport Park. I would guide one of them through the day, while the other would be taught by the Garmin Catalyst. This tablet-like device uses AI, GPS-based data acquisition and accelerometer data to coach drivers to a theoretical best lap. 

Of course, this is not a purely scientific exercise. The only way to accomplish that would be to separate a pair of identical twins at birth and raise them in carefully constructed parallel environments to a suitable age before conducting our experiment, culminating in them fighting to the death. This path, however, was not within our editorial budget or time constraints, and apparently there were also ethical concerns. 

So, fine, we’ll do it the hard way.

Two Meatbags, Coming Up

Our two test subjects came from the GRM staff. Neither had much track time. 

Colin Wood, who manages much of our digital editorial flow and constructs a clever meme or two, had practically zero track experience outside of racing games and sims. Chris Tropea, who runs our video department, has been around the track hundreds of times, although much of that was while hanging out of a truck, camera in hand. 

Since Chris had a little more time at the FIRM, and has heard me dissect the facility lap by lap when discussing data for our track reviews, I felt he had a bit of an advantage going in. As such, we chose him to be the test subject at the mercy of the Catalyst’s digital wizardry, while Colin ended up with me as his personal coach. 

Welcome to class, fellas.

Mars University: Knowledge Brings Fear

We began the day with a classroom session. Actually, this took place well before the test day via video chat so both drivers had some time to digest the information coming their way. We covered all the basics, like looking up and looking ahead, executing smooth, deliberate inputs, relaxing muscles, expanding spatial awareness, and understanding the entry/apex/exit progression. 

At the end, the drivers seemed a bit overwhelmed but also excited. “Obviously these are all terms I deal with a lot in the content we produce,” Colin said, “but being able to connect them to actual theory is nice. None of this means I’m not still super nervous, though.” 

Chris had similar thoughts: “Yeah, I hear you talk about this stuff all the time and I feel like I have a pretty good idea of how it all fits together, but I’ve never had to apply it to something I’m personally going to have to do before.” 

And I told them both that they had 100% the right mental approach. Driving cars fast on track is fun, but the stakes are also high. Enthusiasm should be tempered with caution, but caution should be contrasted with enthusiasm, as a solid sense of healthy confidence is sometimes your best path to success.

Everyone, Suit Up!

We finally arrived at the track, and our plan was thus: Both drivers would get a morning safety briefing followed by taxi laps. Then they’d alternate sessions in one of the FIRM’s NB Miatas. 

After each of Colin’s sessions, he and I would debrief and go over data as well as his thoughts and feelings. When Chris came back, he’d review his laps with the friendly AI who lives inside the Garmin Catalyst. I’d be there to help him with Garmin operation to streamline his workflow, but I’d offer no advice or input aside from device operation.

To show them what the track should look like at speed, early that day I took both of them for a few shotgun laps. “It’s obvious pretty quickly why driving a proper line is so important,” Chris remarked. “Corners that I thought were tight are much more open, and corners I thought were open are practically straights.” Yep, that was the idea.

Colin’s approach was a bit more muted: “It’s definitely more intense than Gran Turismo.” He was learning fast.

Chris drove first and immediately spun in the second corner. I shot a glance at Colin. Just as my head swiveled, Colin wiped the smile off his face.

When Chris got back after that tenuous first session, he quickly reminded me that I had spun a Hyundai Elantra N in the exact same spot. I quietly respected his game and let him go over the Garmin Catalyst’s data. 

Colin’s first laps were slightly faster than Chris’ but also impressively steady, showing a delta of only a couple seconds after an initial out lap for sighting–which was kind of his plan. He remarked that the nervousness hadn’t disappeared, but turning steady laps and building muscle memory was making each subsequent lap more comfortable. I responded that he was already learning a lot, and I’d be in the air-conditioned lounge if he needed me.

When he finally found me in the green room, we looked over his data and debriefed. The big factors slowing him down were the usual: lots of coasting and being too easy on the brakes. Both represented low-hanging fruit for his next session. Soon, I knew, he’d pick up some confidence while shaving seconds. 

Chris’ first data trace actually had a decent overall shape. All those years of watching my “Don’t coast” and “Make decisive control inputs” schtick from behind a camera seemed to have rubbed off on him. While his brake applications were overly gentle and his cornering speeds were low, his overall rhythm looked good. The Garmin Catalyst would theoretically help refine his technique. 

The Catalyst tends to rely on a few main bits of advice to improve lap times. Instructions like “Brake later” or “Apex earlier” are about as in-depth as the advice gets, but when you’re operating a 2500-pound hunk of steel and glass at speed for the first time, that’s plenty. My hope was that those tips would naturally lead him to better technique. Braking later demands harder initiation, for example, but it was up to him to make that leap based on the Catalyst’s advice.

The data traces of our novices’ first laps tell us we have plenty to work with. Chris’ red trace shows that spending a lot of time at the track has clearly rubbed off, however, with much less coasting than Colin (blue trace).

Approaching the Speed of Light

Colin’s second session looked impressively better–clearly due to my expert tutelage–with his speed trace representing that of a more experienced driver. During his debrief, we discussed a lot of specifics, like making decisive brake applications, feeling tire load via the steering wheel, and minimizing the time the car spends doing nothing–and how, if the car’s doing nothing, you’d better be at full throttle. 

His time seemed to reflect this input, dropping close to 13 seconds during those first sessions. More importantly, perhaps, the data showed a more methodical and measured approach.

Comparing Colin’s first lap (blue trace) to his afternoon’s fastest lap (red trace) shows a lot of progression in speed and, perhaps more importantly, smoothness. His transitions are crisp, with minimal speed changes in the corners. 

Chris also started showing steady improvement, and by his third session, he matched Colin’s 13-second drop. Many of Chris’ improvements came via cornering speed; the robotic voice to consistently push harder seemed to be resonating. 

The Catalyst was also urging him to get on the power sooner, and that seemed to be paying off: Chris was now hitting fourth gear on straights where previously he was only getting into third. The Catalyst’s tendency to keep pushing his braking points had also forced him to brake harder, and while his braking traces still needed a bit of polish, the initial cleanup had certainly been done by necessity if not by design.

Chris only worked with the Garmin Catalyst, and its urgings resulted in more aggressiveness later in the day–along with faster speeds. His later speed trace is a bit more ragged, though. Adding smoothness to his inputs could save him even more time.

Good News, Everyone

At the end of the day, both our test subjects loaded up for their final laps. Both had already improved dramatically, so we knew that part of the experiment was a success. At this point, to be honest, I wanted them to relax and enjoy some track time. My only parting advice to Colin: Don’t worry about going fast, just try to do everything right.

He dropped a few more seconds and was now beating his baseline by more than 16 seconds. Best of all, though, his speed traces looked real: smooth transitions, reasonably sharp brake peaks, little variation through the turns, and few signs of forcing the car to do things it didn’t want to do. 

I had pressed the concepts of smoothness and efficiency all day–after all, Colin was in a low-powered car on street tires–and that message seemed to have landed. Yes, there were a few spots where he could have still easily found some time–he showed more confidence when braking from slower speeds than faster speeds–but he had greatly progressed from the stage of “Yeah, there’s lots to work with here.”

Likewise, the Catalyst helped Chris realize significant drops in time–about 18 seconds, in fact. His greatest improvements came in braking and mid-corner speeds, which made sense when the Catalyst was pushing braking points and suggesting better apexes. 

Interestingly, while the speed trace from Chris’ best lap was clearly faster than his baseline lap, it was also clearly not as smooth. There was more variation in long corners where he was learning by trial and error.

Chris’ red trace is a bit rough in spots, showing that he’s making speed with aggression and pushing his limits. Colin’s blue trace is quite smooth, yet he’s still too timid in some corners. The big takeaway here is that both traces look extremely “real.” 

Will Robots Steal Our Jobs?

Both options helped our drivers go faster, but for all of its processing power, the AI option presented a few limitations. “The Catalyst was awesome at telling me what to do and even where to do it, but it never addresses the why and the how of doing those things,” Chris said at the end of the day. “So I just had to trust it and figure it out on the fly.” 

Colin’s experience was a little different: “Sometimes the most useful thing you told me was to take a deep breath and settle down right before I headed out. That made it easier to think about the technical stuff we had gone over, which can feel a little intimidating when you’re just reviewing your session as a squiggly line on an iPad.”

Ultimately, I think these are both viable training methods, but the Catalyst seems to really fall into its wheelhouse when working with an experienced driver who understands the basic techniques and is looking to dial in things. It’s a data analyst, not a coach, and if your skills are at the level where you can take advantage of its analysis, modern data acquisition–no matter the brand–will serve as a very valuable tool.

I think data acquisition is a tool that could help an intermediate driver progress into the advanced group more effectively than it would help a novice driver get into the intermediate group. It will certainly help that novice driver improve lap times, but it might not give them the full context for understanding why things are working the way they are. To fully leverage the AI, you need to bring some existing skills to the table. 

On the other hand, a good coach is going to hold a your hand while you progresss from novice to intermediate to advanced to pro to superpro to living legend and every step in between by using softer, more psychological approaches and better contextualizing those suggestions.

Of course, there are some caveats. That coach is also going to want to be compensated for their time and effort and might not fit your schedule or learning style. (Unless you wait for Black Friday, a Garmin Catalyst is a one-time $999 investment.)

At the end of the day, literally and figuratively, we took two dudes who had never really turned a wheel on track in anger and got them comfortable enough that they were itching to come back as soon as possible. The real excitement is that there’s multiple paths to that enthusiasm.