Hacking your brain: We drive the £3m simulator designed to break your senses
The 2017 Formula show has begun, and new rule changes mean drivers such as Lewis Hamilton and Sebastian Vettel will be racing with cars unlike anything we’ve ever seen. Wider and longer than last year’s cars, this year’s racers will also have larger tyres, giving the drivers unprecedented aerodynamic and mechanical grip.
However, thanks to reams of data fed through highly sophisticated simulators, drivers knew how their cars would handle before a wheel was even turned – and months before engineers they’d hit the track.
In 2017, driving simulators play a huge role in road- and race-car development, and teams can spend millions of pounds on them. But just how realistic are these room-sized simulators, how do they work, and most importantly, what’s it like to drive in one? I visited Ansible Motion, a company that makes and supplies simulators to the world’s largest carmakers and motorsport teams to find out.
Ansible Motion Simulators
Located near Wymondham, Ansible Motion’s base is near to the famous Lotus team, but has none of the notoriety of its neighbour. However, despite its unimposing building, Ansible’s client list includes some of the most well-known racing teams and car manufacturers in the world – all with one need in common.
Ansible Motion is there to solve a problem, says its Technical Director Kia Cammaerts, “the challenge of the driver in the loop”. Simply put, cars don’t exist without humans, and that means that they can’t be developed in isolation, without us. Developing any car involves millions of decisions, and a human will ultimately judge all of them. From the shape of paddles to the behaviour of traction control and semi-autonomous functions, technology that involves the driver needs to be honed and developed by human testing – and that’s where a driver in the loop (DIL) simulator comes in.
In the past, carmakers used countless physical prototypes to get things done, but in 2017 simulators offer a reliable alternative. They’re used for testing physical parts such as gear sticks and buttons in the cabin all the way through to full vehicle performance characterisations, but they’re also crucial for systems where it’s important to have human feedback.
“[Simulation is essential] where what the person driving the car does matters,” says Cammaerts. “Assistive systems try to assist the human, in carrying out what the human wants to do, but what the human is trying to do is affected by how the assistance system works. So there’s a feedback loop there.”
This feedback loop allows engineers to quickly tweak their software and then analyse the data, before going back and inputting new data into the simulator. This feedback process is much faster than doing the tests in real life, and in something like Formula 1, where you have limited time to set up the car every weekend, it’s a massive help. By running through things before a driver even gets to the track, teams can use their track time more effectively.
In the road-car world, systems such as adaptive cruise control are pretty much bulletproof in ideal conditions, and the real test comes when they have to operate in strange, unusual ones. Simulators can set up crucial factors such as weather and road surfaces for freak scenarios – and it’s much safer than testing them in the real world, too. “Systems like AEB [autonomous emergency braking] are hard to test morally in real life – and a simulator represents the quickest, safest way to validate them,” claims Cammaerts.
But while simulators seem great, there is one issue with them. Real-world testing is perfect testing, so why use simulators? Testing in the real world gives you an undoubtedly true handle on how products will work when they hit the road, but it’s often hard to set up the exact situations you want to test. Simulators, on the other hand, gives engineers total control of the environment from a physics perspective, so they know the data that comes out is 100% related to the data that went in – with no unforeseen external factors.
Getting the realism
While simulators may represent the most cost-effective option, the data they’d produce would be worthless if they weren’t convincing for the humans participating in the virtual test-drive. To achieve proper feedback from a human’s brain, you need to trick their senses into thinking they’re in an actual car.
“Will the person operating a [conventional simulator] behave in the same way they would in a car? The answer in almost all cases is no,” says Cammaerts. “So if you’re developing an assistive system… it needs to be immersive.” And that’s where Ansible Motion’s simulator sets itself apart from the crowd.
The Delta series DIL simulator I’m about to use takes up a large room, but its main component, the sled, is smaller and less powerful than you’d expect. Rather than a capsule supported by hydraulic joints and servos, Ansible Motion’s simulator is actually pretty small.
It consists of three main parts. A car cabin “sled” sits on top of a mixture of cables and framework, while images are fed through screens eight metres in diameter that wrap themselves 270 degrees around the room. At the same time, racks of 24 computers manipulate the sights and sounds I’m about to experience, but that’s pretty much it. From the outside at least, it doesn’t seem to have the power to generate the G-forces you’d expect from a simulator that can train F1 drivers – but then it’s not supposed to.
The Ansible Motion simulator hacks something called the human vestibular system, and tricks your senses into thinking you’re moving. Each part of the simulator is designed to maintain the illusion of motion, and it features numerous devices designed to basically hack your head.
The main targets are your semicircular canals and otoliths, areas of your inner ear that help you detect linear motion and rotation. These organs are in charge of motion and spatial orientation – like the feeling of flying forward when you brake, or leaning to the side when you go around a corner, for example. And almost all the features of the Ansible Motion sled are designed to target them.
To do this, the Ansible Motion rig does something called motion cueing. It gives you the beginning of what you’d feel when braking or steering in a car, and then uses the visual part of the simulation to make your brain almost assume the rest. The result? The simulator is able to simulate nuanced and rapid direction changes, and provide a real sense of vehicle stability – or instability – unlike any entertainment experience you’d see at funfair.
Trying the simulator
After observing one of the simulator’s test drivers on the rig, it’s my turn to drive the simulator. Because of the way in which the simulator motion-cues – using visuals to follow through suggestions of motion – watching from outside the rig is actually quite nauseating. I’m told that these seemingly unsteady visuals are all part of the way the simulator hacks its occupant’s senses, and I’ll be fine once I’m in the driver’s seat myself.
Even without the software running, the simulator is pretty convincing. With a roof over my head, and the same fittings as a real car, this simulator is miles away from the gaming rig I’m used to. Once I’m settled in, an engineer talks to me through the headphones I’m wearing. First I’ll be driving through cones to acclimatise, and then I’ll move onto high-speed circuit driving, at Spa.
As soon as the rig is primed and the software is loaded, it feels like being in a real car. The graphics are no match for something such as Project Cars on a high-end PC, but every window I look out of features the same virtual environment, and because it wraps around the screen, it’s in my peripheral vision too.
The simulator packs in so much detail, it’s possible to take a lot of it for granted. The rear-view mirrors and wing mirrors feature the same environment, so I can’t help but feel you’re actually there. The cabin is also filled with low vibrations, designed to simulate a running engine.
Once I’ve acclimatised to the general controls, and brake and throttle response, Spa is loaded up, and I find myself in the pits. Pulling out of my garage just feels real, and as I build up speed for Eau Rouge – one of the best corners in the world – I feel like I’m there. My car dances over every bump, and as I hit apexes and drift wide onto the rumble strips, I can feel what the car’s doing through my seat, but also my head.
After getting the hang of the front-wheel-drive car, the engineer tells me it’s time to move onto a more powerful rear-wheel-drive car, and that’s where the fun really begins. It’s louder, and feels faster – but it’s also tail happy, and and after a while I start learning to drift it.
When carrying a little too much speed into the corner, all it takes is a sharp turn and you can feel the tyres lose traction – and then it’s just about using the throttle to keep the slide going. The whole time, I’m almost convinced there’s a car beneath me, as all my senses – from my sight to my balance – tell me I’m driving.
The most convincing moment comes later on in the session. After getting more comfortable, I start braking later, taking more risks and generally going for it. But at one corner, I brake far too late, and it suddenly dawns on me I might fly off the track. At that point, under heavy braking, with my body pulled forward and the wall approaching, I realise I’m actually scared of crashing.
It’s the same feeling from your gut you get when braking to avoid an accident, and the fact I feel it in the simulator is pretty amazing – and terrifying. It’s strange, knowing something isn’t real but not being able to stop my body reacting to it, and it reminds me of my first experience of VR.
After a few more laps, the engineer comes on the radio and asks if I want to try a final, “nuts” car. It’s an ALMS prototype that’s essentially a Le Mans endurance car, and he tells me it’s pretty brutal.
When companies buy these simulators, they often ship out their best drivers to test them beforehand, just to make sure they correlate. And with a client list including huge F1 teams and three of the top five car companies in the world, I realise this is as close to driving a racing car as I’ve got.
Driving a racing car
According to Cammaerts, the ALMS isn’t a car most people get to experience in the simulator, since it’s so aggressive and inaccessible compared to what’s come before. And as the simulation loads again and I pull out of the pits, I can feel what he’s talking about.
The ALMS car transmits every single bump in the road, because there’s not even a hint of suspension give. Huge levels of downforce cause G-forces to throw me around on faster corners, and their absence makes the car skittish on slow ones – but the biggest change comes from the braking. There’s pretty much no travel in the pedal, and even with a lot of force the pedal isn’t moving. However, I’m assured by the engineer that it just requires huge amounts of force to stop the car.
Put these factors together and driving the ALMS car is a pretty heated workout. Wrestling the steering wheel around faster corners, I feel like a ragdoll inside the car, and even the simple act of braking feels like I’m doing a some sort of leg exercise. About ten laps later, it’s time to stop and I’m pretty tired to be honest. Cammaerts tells me that Ansible Motions also offers a special helmet that increases the G-force and neck-load experience, but I’m pretty sure that would’ve killed me.
This simulator is clearly great for motorsport, but it’s clear that in the coming years simulators like Ansible Motion’s will play a huge role in the cars we drive in the future. Systems such as semi-autonomous functions and driverless tech need to be tested with the inputs of humans, and machines such as Ansible Motion’s DIL simulator should make our transition from manual driving to driverless systems safer, easier and more seamless.
- Simulator in the R&D Centre runs on 20 computers plus four audio systems
- The system uses five projectors that cost £20,000 each
- A standard Delta DIL simulator like the one I used costs £2 million
- The Spa track I drove is a 6GB file
- The car I drove is only a 70MB file
- The simulator weighs roughly two tonnes, with a cabin of approximately 500kg + me
- The stratiform construction of the simulator allows for six degrees of freedom of movement