Posted by Mark DeShon

      If you’ve been around autocross for any length of time, you’ve likely heard the word ‘data’ thrown around the paddock quite a bit. ‘Did you look at the data?’ ‘What did the data say?’ ‘The data said my co-driver is an alien who can teleport himself to the next element in record time.’ Data analysis is currently the most modern way to share the secrets of speed between drivers.

     If you’ve wondered what exactly is all this data that people are talking about, fear not, I can explain in one sentence. It is what you make of it. Okay, so that wasn’t exactly a thorough explanation, but it does capture the essence of the expansive capabilities of the technology that we’re talking about. In this article we’re going to scrape the surface of some of these capabilities. While GPS data has its roots in Formula 1 and other forms of road racing, today we’re going to focus on its application in autocross. Also, as data can be used extensively for car diagnostics and tuning, we’re going to limit the conversation to its driver development capabilities for now.

     For starters, the ‘data’ we are referring to is GPS data which is run through special analysis software. This software comes in many forms with varying capabilities and user interfaces. The basis of the technology is a GPS unit which logs car position  on course during a run and records those positions with time stamps. This usually happens at a rate of between 5Hz and 20Hz, or GPS points per second. These points are then fed to the analysis software which plots them onto a course map and what is called the ‘speed vs distance graph’.

The speed vs distance graph is at the core of why we use GPS data and gets right to the point; showing you your speed at any point over the distance of the course (thus, speed vs distance). Users usually must define the start and end points of the course, and can also break the course down into sectors. When compared with multiple runs, users can readily identify their strong points and weak points in any given run when compared to a baseline run. At any point on the speed/distance graph, there will be a continuous “time gap”, which shows how much further ahead or behind the given run is when compared to the baseline run in seconds.

A screen shot from the Race Trace iPhone App showing the speed plots of the baseline run (purple) and the analysis run (green) and the time gap (shaded). Note the .8 second shift in time gap near center screen caused by the green driver’s eagerness with the throttle; giving very little gain in the short term, but putting him behind and costing him nearly 15 MPH  by the end of the transitional section... I'm the green driver.  Photo: Jeff Stuart 

     Data can be helpful for a single driver comparing multiple runs on the same course, especially during a practice day. It helps when experimenting with line selection, especially in give-and-take type compound elements; giving yourself varied data to compare. Of course, this could waste valuable runs if attempted in competition, but can be very useful for identifying and retraining bad habits in the long term.

     However, if you want to reap the most benefit from data analysis, the best way is to compare your data to that of another driver in the same car. Having a full-time codriver with whom to share secrets and compare data is most often beneficial to both parties. Even if one driver is consistently faster than the other, it is not uncommon for the faster driver to identify certain sectors or types of elements where the over-all slower driver has the upper hand and learn from them. Even championship winning, top-tier drivers have been known to put fast guest drivers in their own car simply to acquire data on what approaches the guest driver uses and to look for potential gains. It’s one of the keys to staying on top.
     Of course collecting the data is only the first step, and GPS data is only as good as the person analyzing it. Again, it is what you make of it. So what are we looking for when we analyze data? Of course, the end goal is to identify areas that lower our total time on course, so we start with the time gap plot. It is important to not look at any one corner or element under too much of a magnifying glass right away, but instead to look at the broader scope of how one element affects the next. For example, a cursory glance of a certain sector may reveal that driver A was .2 seconds faster in sweeper X, but driver B was .5” faster in straightaway Y which followed it. This likely does not mean that driver A is better at sweepers and driver B is better at mashing the go pedal, but more likely that driver A over-drove sweeper X, and sacrificed his corner exit speed, resulting in a .3” net loss by the end of the next straightaway.

     Autocross courses are often an intricate mix of compound elements, but with experience can usually be broken down into a handful of key pinch points and setup opportunities which, when attacked properly will allow maximum full-throttle time within the intermediary elements. Most often you will find that the biggest losses of time when compared to a baseline will correspond with an area where the baseline driver was able to stay full throttle for a longer distance on course. Identify such an area using time gap, and look at what he or she did differently in the element preceding it using speed vs. distance.

Two screen shots from Solo Storm showing the green driver leading the red driver by .029 seconds going into the final corner (left), but entering too hot and pushing wide putting him .215 seconds behind by corner exit (below).
Photos: Leonard Cachola 

     Analyzing the time gap will give most drivers volumes of information to work with and apply as they continue to hone their driving skills. Yet data systems can get much more complex and expose strengths and weaknesses in a driver’s style far more nuanced than simply corner speeds and line selection. Most data systems can tap into the car’s ECU via OBD II or CAN ports and read things like throttle position, individual wheel speeds, steering angle, and brake light switch. They can also accept stand-alone sensors like oil pressure or temperature, or brake pedal pressure. These finer points can help you isolate which problems are car setup induced and which are driver induced and how the two relate to each other.

     Vehicle sensor data can also reveal minute differences in driving styles which have big impacts. For example, you can have two drivers braking at the same point for an element and slowing to the same speed initially, but one driver struggles and slides through the element while the baseline driver cruises through effortlessly. Looking at the speed at which each driver releases the brake pedal (pedal pressure change), their speed of turn-in (steering angle change), as well as the overlap of the two can offer the answer as to why one driver is upsetting the car and the other isn’t when both are attempting to corner at the same speed.

     When setting up the more complex data systems, usually any raw input can be compared to any others via mathematical formulas and graphed. Many of these formulas are available on the internet or can be derived from scratch. This can be helpful to further isolate the more subtle differences in driving style. For example, yaw rate can be mathematically compared with steering angle and be plotted as an understeer vs oversteer graph. This can help identify how much slip angle each driver demands from the front or rear tires and, when compared with the time gap, how well that’s working for them. It will allow you to see things like how long the driver stays within the most effective slip angle, how often they ‘hunt’ for more grip mid-corner, and how good their reaction times are when correcting an oversteer moment.

     This is just one example. Practically speaking, there are endless combinations of inputs that can be processed and displayed graphically for mental digestion. As I mentioned, it is what you make of it, and the software is only as good as the person setting it up. 

 The above DL-1 data log shows Lateral G (top) and Understeer/Oversteer (bottom) plots as well as a course map. At the cursor point you can see a spike on the understeer graph. The corresponding spot on the Lateral G graphs shows peak Gs cropped at around 1.2G, whereas other corners on course are seeing peak Gs of upwards of 1.4G. This is a glimpse into more advanced, modular data software. Photo: Marshall Grice

     As you can surely tell by now, the possibilities are endless and can be a little daunting, but a nice thing about data analysis systems is that they can be set up to focus on a certain driver’s needs and goals. If all you want to do is look at time gap and call it a day, you can purchase a phone app for less than $20, pair it with a $100 GPS unit and go find some more speed. If you want to spend late nights completely invading the privacy of your vehicle’s systems to understand where communication breaks down between you and it, $1500+ systems are out there to help you do just that.

     It is what you make of it, but there are benefits to users at all levels of data analysis. Drivers have been verbally sharing secrets of speed ever since the automobile has existed, and data analysis is simply the next iteration of that. This powerful tool can open the eyes of even elite drivers and answer the ‘how’ question that is so often hidden behind the time slip. 


Leave a comment

All blog comments are checked prior to publishing