In this current era of Formula One (and other top motorsport series) Driver-in-the-Loop (DIL) simulators have become crucial development and preparation tools.
Not so long ago, DIL simulators were primarily used as supplemental devices – mainly for drivers’ circuit familiarisation and basic setup work. But this has changed dramatically.
With tighter track testing restrictions and increasingly complex cars – as well as advancements in various simulation technologies – DIL simulators have been reclassified as essential tools, and are now regarded by top teams as controlled, repeatable virtual testing environments that can mirror real racing scenarios in order to assess new aerodynamic packages, optimise vehicle setups and evaluate race strategies – all in advance of and in parallel with real cars turning wheels on race tracks.
But achieving this utility is easier said than done. Successfully supporting a racing programme with DIL simulation requires procedural diligence and attention to details that are somewhat outside the box.
Getting it right delivers huge rewards. Getting it wrong can be costly.
Adrian Newey recently joined Aston Martin F1
In recent articles, Adrian Newey – who has signed onto a 5-year contract as managing technical partner with the Aston Martin Aramco Formula One Team to help it navigate the new F1 rules era that is already unfolding – has been openly speaking about weaknesses in the team’s simulator programme. His comments are revealing.
Newey said to the media:
I think it is fair to say that some of our tools are weak, particularly the driver in the loop simulator. It needs a lot of work because it's not correlating at all at the moment, which is a fundamental research tool. Not having that is a limitation. But we've just got to work around it in the meantime and then sort out a plan to get it to where it needs to be. But that's probably a two-year project in truth. So, we're going to be a bit blind on that for some time. We have just got to try to use experience and best judgement. How successful that will be, time will tell.
In addition, Newey voiced the following about DIL simulation on the Aston Martin F1 website:
You can have the best motion system in the world, but if you don’t have the modelling to go with it, and correlation with the aero model, correlation with the tyre model and so on, it won’t be of any use.
Responding to Newey’s critique, team principal, Andy Cowell, confirmed that the team will require time to optimise the DIL simulation tools, stating:
I think whenever you create new equipment, it takes a while to commission it and then work out how well correlated either the DLS [Dynamic Lap Simulation] or the wind tunnel is with the real world. So, you need to do some updates, bring it to the track, you need to see how everything matches up. That’s the correlation. And even experienced teams have problems with correlation. You hear it up and down the pit lane.
From other media sources, it’s apparent that Aston Martin F1 is taking immediate action to course correct its simulator programme. Some of this involves a changing of the guard such as signing Newey’s former Red Bull colleagues, Giles Wood and Gioacchino Vino to help improve the team’s simulation tools. (Interestingly, Wood, who has joined Aston Martin F1 as simulation and vehicle modelling director, has been lured back into the world of F1 from his most recent role at Apple, where he was involved with autonomous vehicle developments.)
In speaking about Aston Martin F1’s simulator, Newey and Cowell are referring to the new simulator that was commissioned at the team’s factory at Silverstone in 2024. We’re not entirely sure what type of DIL simulator this might be – although we can make an educated guess. We can only say that it is not an Ansible Motion simulator. And it pains us to hear comments like Newey’s because we understand from experience that the team is facing a bigger challenge than it may realise.
As we’ve discussed in previous articles such as our Driving Simulator Myths series, it seems that many early-stage assessments of DIL simulators focus almost exclusively on simulator motion capabilities. Newey’s comment about “You can have the best motion system in the world . . .” speaks directly to this point, and it may indicate that the team fell into this subtle trap prior to Newey’s arrival.
Motion machinery is certainly a visually dominant part of a driving simulator installation, and we know that automotive engineers naturally gravitate toward motion systems because they are more “relatable” than many of the other DIL simulator systems. But we would assert the following: While it is certainly important, motion capabilities do not singularly define the quality of a driving simulator or its capacity for correlation.
Driver-in-the-Loop simulation: A multi-sensory experience
DIL immersion is a multi-sensory endeavour, an orchestra, and motion is just one of the players. In other words: Motion is but one sub-category of one type of sensory stimulation that addresses just one category of human senses. If we stand back to look at DIL simulators from a big picture perspective, across the entire spectrum of correlation requirements, is motion cueing any more or less important than vision cueing or auditory cueing or any other type of sensory cueing? Probably not. All of these sensory triggers must work together seamlessly, in a convincing way.
In addition, it’s worth noting that the purpose of a driving simulator’s motion machinery is not to move a car around or to replicate the motions of a real car (as touched upon briefly in the previous article titled, Driving Simulator Cueing Fundamentals). The purpose of the motion machinery is to move a human being around, to convince them via inertial stimulation that they are interacting with a real car. This is an entirely different task.
DIL simulator correlation – which Newey correctly identifies as the cornerstone to making progress – can therefore be a trickier proposition than might first be imagined. It starts by viewing the simulator as a whole – not just as a motion machinery kit – in the context of measurable driver interactions. We speak from experience on this, as we’ve been tackling it for over 15 years.
Ansible Motion Delta S3: Preparing for a virtual test drive
Rather than continuing to speak in generalities, let’s get into this a bit by addressing some of the logical steps and technical challenges that are faced when aspiring to use DIL simulation with confidence in motorsport applications such as F1, as a correlated, trusted tool.
Let’s say that the ultimate goal is to have an operational DIL simulator laboratory that provides useful race strategy insights for key areas such as tyre degradation and energy management. And let’s also say that we want the ability to explore components/setups such as front wing stiffness variations.
We know that, just as is the case with real testing, there will be two primary streams of information that can be used to guide the simulator (virtual testing) programme: objective data via telemetry and subjective driver feedback. And we know that DIL correlation relies upon comparisons. But what should be compared?
On the telemetry side alone, there will be several sub-categories of available information. Most notably, there will be (1) data from the simulator itself that can be used to understand and define such things as the motion and vision cueing settings – and which is, notably, a bit of a data island because it will have no real world comparison reference unless the team operates multiple DIL simulators; (2) data from the vehicle physics model that powers the simulation, that can be directly compared to real vehicle data (such as vehicle speed, lateral acceleration, and so on); (3) data from the vehicle physics model that powers the simulation, that cannot be directly compared to real vehicle data (such as tyre slip angles, individual component loads, and so on); and (4) supplemental data related to driver biometrics.
The above information, when properly categorised, adds another layer to the already significant data analysis and interpretation load for race engineers. And some of the information, as noted, does not map directly to traditional on-car data acquisition and analysis. So, Newey’s comment above (“. . . if you don’t have the modelling . . . and correlation with the aero model, correlation with the tyre model and so on, it won’t be of any use.”) is really just the tip of an iceberg that addresses baseline physics model correlation, but may not address correlating DIL simulator performance to meet driver expectations. For example, once it’s implemented in a DIL simulator, a “correlated” tyre model can fall apart if its interactions with the real-time track surface model is not fully sorted.
Ultimately, it is vital that a DIL simulation (consisting of a virtual car operating in a virtual environment) is conveying correct control and stability information to the driver, and that the driver is behaving in ways that mirror interactions with a real car (steering corrections, braking and throttle applications, orienting apexes, establishing lines of sight, etc.)
Ansible Motion Delta S3: Simulating track surface variations [video]
We’ve explored a number of interesting data analysis and correlation techniques over the last decade and a half to help our customers around the world leverage their DIL simulators for on-track success. Here are a few approaches that may be helpful for race teams that are in the midst of tackling this challenge:
The truth is as cited above: “Even experienced teams have problems with correlation. You hear it up and down the pit lane.” And it’s also true that correlation becomes a multi-layered endeavour when DIL simulators enter into the mix. As a DIL simulator manufacturer, we are enthusiastic participants in the quest for perfection. Using our accumulated knowledge and field experience, we often work closely with our customers to help them make the most out of their simulator installations. Winning virtual races may not be as exciting as the real thing. But when our customers use our simulators to achieve on-track success, we’ll confess that it feels pretty good.
If you are interested in reading more about the current state of the art in motorsports driving simulators, we invite you to download our FREE eBook, Engineering the advantage: Driver-in-the-Loop simulation in motorsport: