The Virtual Reality of Racing Simulators

You are hurtling down the frontstretch at Michigan, your speed approaching 215 mph.  Your seat moves up and down as you hit the seams, but your focus is squarely on getting into Turn 1 losing as little speed as possible.  You squeeze the brakes and feel yourself moving forward, only to realize that you’re still moving too quickly. As the car starts to head toward the wall, you panic and squeeze the brake even harder.

The car snaps loose and the next thing you feel… is an engineer’s hand on your shoulder.  You turn around to see her barely suppressing a smile.

“Let’s try that again. Maybe you want to brake earlier this time, huh?”

The latest racing simulators are far more involved than your steering-wheel-feedback-enabled video game. When you hear drivers talk about using simulators to familiarize themselves with new tracks, you are only hearing about the surface layer.

Ford opened a new Racing Technology Center in Concord NC just about a month ago.  The old Ford facility used to be called “The Shack” because of it’s size. The new facility, which is across the street from Roush Fenway Racing, is 33,000 square feet. One of the main features is a driving simulator similar to the ones that F1 uses.  Five screens provide a 180-degree view for the driver. The cockpit is the front half of a NASCAR Sprint Cup car that is set on a full-motion platform.  The platform duplicates the exact motions you would feel in the car – bumps, slips, sways, yaw.

Let’s move away from the driver for a moment, because he (or she) is a relatively recent addition. All race teams use vehicle dynamics simulations: computer programs that predict lap times for specific setups. A crew chief changes the setup on the computer – camber, cross weight, track bar, etc. – and can see in advance whether the changes make the lap time better or worse.

Simulation programs have to be validated, meaning you have to make sure they correspond with reality. Every team also has (or rents time on)  K&C and seven-post rigs.  (K&C stands for Kinematics and Compliance – I’ll be getting into those in upcoming posts.) These machines attempt to quantify how the car responds to external changes, like turning, bumps, etc.

Engineers thus went back and forth between theory (the simulations) and experiment, developing a model of the car, testing it against how the car behaved, and then refining their model.  (Unsurprisingly, this is exactly how scientific research on things like alternative energy sources and cancer works.)

This is a great model for the Google driverless car. But that’s not how racing works. Racing requires a living, breathing, thinking (hopefully) human being in the seat who has to constantly take in information, process that information and act on it.

And that’s where racing simulators are moving. The buzzword is “Driver in the Loop”, which means that you’re creating a model that includes the driver.  This is not an easy step. Drivers are very different in terms of their preferences for set-up, what they’re comfortable driving, how loose they’re willing to be early in a run to make the car faster later, etc.

The simulator in the new Ford Tech Center is a sled-type simulator. Less advanced models have hydraulic pistons that raise and lower the cockpit to simulate bumps and change attitude.  The sled can actually duplicate all six types of motion: three linear motions (up/down, left/right, front/back) and three rotational degrees of freedom (yaw, roll and pitch).  This motion platform was developed by a British company called Ansible Motion and the picture below is from their website. You can see the sled rails at the back, and the 180+degree surround on which the images of the track are displayed. The steering wheel provides feedback to the driver and even the seatbelts are cued in so that when you brake, the seatbelt tightens just as it would in a real car. Ansible is the same vendor that worked on McLaren’s F1 simulator. The Ford folks claim that this design has a much faster response time, meaning that the time between when you turn the steering wheel and when you feel the result, is shorter and more like real life.

At present, they’ve got ten tracks in the library for the new simulator – eventually they will have all the NASCAR tracks and, since the Tech Center is meant to support all of Ford’s motorsport activities, they will be able to change out the cockpit to, say, a Daytona Prototype, and including tracks like Sebring.

So far, I’ve made it sound like they’re just one-upping iRacing, but a prime feature of the center is that there is a whole room associated with the simulator that is filled with engineers who are watching both the driver and the racecar input/output data.  The driver is being assimilated into the simulations. This is why it’s called “Driver-in-the-Loop” simulation.

On the one hand, the driver will help validate all the models.  If there’s a bump on Pit Road at a track that gives a driver trouble, he or she will recognize when it’s not in the model of the track used by the simulator. (And since tracks change significantly, the models have to change to keep up with reality.)

While we’re talking tracks, let’s point out that the track models aren’t made by someone sent out with a tape measure. Tracks are laser scanned to a resolution of a few millimeters. Every dip and bump is recorded and used in modeling the tracks. Laser scanning not only collects three-dimensional location information, it looks at the quality of the reflection.  Laser scanning can differentiate between a white painted line and a yellow painted line, between two different lanes of asphalt, or even skid marks. When you’re traveling at top speed, any surface irregularity becomes important because all it takes is for the car to be throw a little out of equilibrium and you’re in the wall.

In addition to the track, the driver can also provide feedback about whether the “car” responds the way it does in real life. But at the same time the driver is evaluating the simulator elements, the engineers are evaluating the driver. They can start to look at things like how a particular driver’s comfort level may dictate a different line for them relative to another driver. They can run repeated tests to find out how constant a driver is. Do they brake the same way going into Turn 1 at Charlotte every time? Or are they hyper aware of something like tire fall off and able to tailor their braking to the condition of the car?  This type of research has great potential to improve communication between the team and the driver.

Skeptics will worry that we’re getting uncomfortably close to a situation in which we have a bunch of engineers sitting around driving the racecar via remote control and the driver is no more than a warm body executing commands as he’s told. The beautiful thing about human beings is that you can’t model a human being. Having done research in both physics and science education, there’s a huge difference between measuring electrons and measuring people. You kick an electron twenty times and it will pretty much do the same thing each time. You kick a person twenty times and (aside from the danger of being kicked back), you’ll get at least ten responses depending on the person’s mood.

It’ll be interesting to see whether tools like this can help the Ford teams (especially Roush Fenway Racing) catch up to Chevy.