Reading Time: 6 minutes

The gap between video games and road cars has, ostensibly, been shrinking for years.

However, while the possibility of a skilled video game driver becoming – or already being – good at driving in real life has been debated, there is one area where the virtual world has provided some very real benefits to road cars.

rFactor 2, a well-known race car game, started life as a simulator for F1 teams. Now, that same tech – though known as rFpro – is being used to develop autonomous cars. Auto Futures spoke to Peter Daley, the Managing Director of rFpro to find out more.

Can you explain the relationship between rFpro and rFactor 2?

rFactor 2 was developed and released in 2012 by Image Space Incorporated (ISI) as a sequel to its popular rFactor computer game initially released in 2005. Rights to sell and develop rFactor 2 were later acquired by its current owner Studio 397.Peter Daley Managing Director Rfpro

ISI had been formed in 1992 to develop vehicle simulation software and the simulation environment around it. This technology and knowledge were used in various simulation projects and to create a number of products from driver training simulators to computer games.

rFpro was conceived in 2007 from one such project in which vehicle simulation work was carried out for an F1 team. The ISI technology used was licensed and rFpro was formed as a separate company to develop and provide specialised, professional simulation software, initially for use in high-end driver-in-the-loop (DIL) simulations. All the relevant technology is now owned by rFpro and a core group of developers from ISI joined the business in 2017.

rFpro continues to be developed as a highly accurate, engineering-grade, DIL, software-in-the-loop (SIL) and hardware-in-the-loop (HIL) simulation environment for vehicle manufacturers, automotive industry suppliers and motorsport teams. It is not related to rFactor 2 except through our common history.

What benefits does testing in rFpro offer to autonomous driving companies?

The majority of miles driven in a car are relatively uneventful but autonomous vehicles need to be subjected to all manner of scenarios, and multiple variations of those scenarios, to be properly trained and tested. It simply isn’t feasible, viable or safe to do the amount of driving required in the real world to cover all the bases.

Using a simulated environment, autonomous vehicle developers are able to train, test and validate their vehicles. Driving simulations create virtual data sets for the vehicle’s AI to learn from, enabling developers to push their systems to the limit in a safe environment. Using rFpro, the full toolchain (sensors, vehicle, control software etc) can be modelled and simulated. Real-world hardware such as sensors and ECUs can be linked with and interact directly with the simulated world.

Simulations can be hugely scaled, limited only by the available computing power, enabling millions of highly valuable miles to be driven in a matter of days.

How does rFpro ensure that its driving scenarios are accurate to real-world conditions?

To create our virtual models, we use kinetic LiDAR surveys to map the road surface to an accuracy of 1mm in the vertical direction. Every ripple, bump, drain cover and road marking is precisely replicated, providing a level of detail that is critical for vehicle dynamic and ADAS development.

For DIL testing, realism and immersion are key priorities for useful simulation. By focussing on the speed of response and frame rate, coupled with high video resolution, the driver receives essential visual cues at precisely the right time.

Tokyo Digital Twin With Wet Road

For HIL and SIL simulations, flexibility, connectivity and realism are all critical. Our interfaces allow rFpro to connect to a range of other hardware and software components that our customers wish to “plug in” to the virtual environment that we provide. The quality and detail in our digital models of the real world are unsurpassed and, if the application requires it, we can run in real-time or synchro-step mode – the latter allowing simulations to be slowed down or sped up to match data processing or transmission limitations.

In all cases, we can vary lighting, time of day, weather conditions and road surface friction in order to explore the full range of real-world conditions for any given scenario.

What results do you give to carmakers to help aid their development?

Our virtual environment can be used to develop and test ADAS and autonomous vehicle systems, vehicle dynamics and powertrain models, or to conduct human factor studies. It is a highly accurate, engineering-grade simulation tool. It gives carmakers the ability to explore, assess, and test their designs far more extensively, quickly, cost-effectively and safely than would be possible working only in the real world.

It is important to note that whilst the power of simulation enables tests to be hugely scaled, correlation and calibration against real-world testing is essential to ensure the simulation is accurate. For example, after a race weekend, our Formula 1 customers will drive the digital twin of the track using their DIL simulators to compare data recorded during the event. Likewise, our vehicle manufacturer customers will use our library of proving ground digital twins to conduct virtual testing and correlation before and after visiting the real location.

You have scanned real city streets in Tokyo and Paris, how difficult is it to mirror these environments in the simulator?

rFpro has built hundreds of kilometres of digital public roads across China, USA, Canada and Europe, including urban city centres, rural routes, highways and motorways. We also have a library of the world’s best-known race tracks used by Formula 1, Formula E, WEC, Indycar and NASCAR teams.


Once we have the LiDAR data and images from a scan we have a team of highly skilled project engineers and talented digital artists to process the data and build the models into realistic and usable replicas of real-world environments. This can take from a few weeks to many months, depending on the size and complexity of the model. We ensure that everything is geometrically accurate and physically define the objects’ materials so that light is reflected or absorbed as it would be in the real world, which is very important for autonomous vehicle sensor development.

The sun and moon are correctly positioned in the sky for the time of year and time of day, given the longitude and latitude of the location, casting accurate shadows over each 24-hour cycle.

Do you see more car development moving to simulators and computers in the years to come?

Yes, we do. Vehicle manufacturers are offering a wider range of products than ever before, the technologies used are more complex and the number of type approval tests required is increasing. This means the amount of effort needed to develop or update a vehicle is growing exponentially.

Simulation provides a means to accelerate vehicle development while significantly reducing costs. Whilst real-world tests are likely always to be needed for final validation, a large and increasing proportion of development and testing can be done using simulation of one form or another.

Are there any effects or phenomena that occur when driving that simulators can’t, or can’t currently, model?

One of the limitations of current generation DIL simulators is that they cannot subject the driver to sustained g-forces. The ability to do this is often limited by the simulator’s range of travel. This is not an issue for 99% of driving scenarios but if, for example, you were to drive continuously around a roundabout the simulator would not be able to sustainably deliver that feeling of cornering.

What does the future of driving look like to you in the next five to ten years?

We work with seven of the ten largest vehicle manufacturers in the world who are in the process of developing next-generation vehicles. The future is clearly electric; customers are using rFpro to frontload the development of electrified powertrains, enabling them to quickly assess multiple motor and battery combinations to find the right trade-off.

There is also a clear drive towards autonomy and automated systems. We believe there will be a growing number of vehicles with level 4/5 autonomy on the market in the latter part of this time frame. However, we expect significantly more growth initially in the use of level 2+ systems as stepping stones towards full autonomy.

Leave a Comment