For a Formula One driver at the top of his profession, 225 ms per lap can mean the difference between success and failure; for an average driver travelling at 120 km/h (75 mph), a 225 ms reaction delay in response to a looming emergency can mean the difference between safety and disaster; and for the operator of a vehicle simulator, a 225 ms lag in input and output effect, can make a nonsense of dynamics development work.
“Imagine turning the steering wheel of your car and then waiting almost a quarter of a second for the vehicle to respond,” says Chris Hoyle, Technical Director of the U.K. software company rFactor Pro. “It makes realistic simulation of vehicle control very difficult.”
In the world of advanced technology simulators, that delay of up to 225 ms, is called latency, and it can drastically reduce the effectiveness of driver-in-the-loop (DIL) systems used for studying vehicle dynamics.
Under some circumstances, it can even cause simulator drivers to suffer motion sickness. But latency is a typical drawback of a conventional simulator, says Hoyle.
In top-notch motorsport, such negative effects are unacceptable. So rFactor Pro developed software for a major race team to overcome the problem. Following that initial application, the software was released to other race teams and, says Hoyle, in its zone of speciality, currently “dominates” the simulator market in F1 and NASCAR.
Now it is cascading into the development of road cars and is set to make a very significant contribution to simulator effectiveness in Europe, the U.S., and Far East, among OEMs, Tier 1 suppliers, and engineering consultancies, reveals Hoyle.
One of these is the Austrian headquartered Tier 1 supplier of drivetrain solutions and testing equipment, AVL. Hoyle states: “AVL estimates that over 30% of the costs incurred in developing driving attributes can be saved by frontloading the engineering activity on a DIL simulator with subjective feedback. It also recognizes unprecedented opportunities for the development of stability systems in a safe and repeatable environment, making the evaluation of extreme maneuvers much more scientific and less hazardous.”
At the most recent Driving Simulation Conference in Paris, Renault’s Zhou Fang, presented a paper describing the company’s experiments in studying stability control on its simulator. The paper highlighted the challenges of studying vehicle dynamics on a simulator with 200 ms of latency, says Hoyle: “Delays make the simulation of driveline events equally difficult. We hear poor gearshifts and driveline backlash as much as we feel them, but the typical audio feedback on a conventional simulator takes 100 ms, which means the sounds arrive out of synchronization with the events, making the study of refinement issues impossible.
“It’s like watching a film where the soundtrack is out of sync with the actors’ lips!”
Hoyle states that even the graphics of most simulators are compromised, producing optically approximated images that can lead to conflicting signals reaching the driver’s brain through the eyes and the inner ear, which impair spatial awareness and can produce motion sickness.
“Until now, all these limitations have confined simulators to ergonomics, human factors and man-machine-interface applications, where the delayed response was not an issue,” he says. “When we set out with the backing of a high budget F1 team to achieve DIL simulation of vehicle dynamics, it meant creating a simulator that was not only physically accurate, but which was also visually and aurally realistic. We developed new software that provides video signals 10 times faster, audio signals 20 times faster, and positions every pixel on the video screens to a fraction of a degree.”
He regards the potential benefits for the mass production automotive industry users of tackling the simulator latency challenge as being greater than those for the motorsport industry: “By supplementing engineering data with subjective ‘feel’ in the early concept stages of a new vehicle program years before real vehicle hardware is available to test, better decision making becomes possible, avoiding costly revisions later.”
The pace of hardware development in the computer industry has helped to make rFactor Pro’s low-latency vision system possible.
To improve response speed and image quality required the use of markedly higher bandwidths than usual for the transmission of video and road surface information.
Hoyle believes his company to be the only supplier to create a bespoke video pipeline to reduce latency, greatly out-performing what he terms traditional systems, to deliver high quality video at 1920x1200 or 2560x1600 pixels at 120 Hz across multiple projection channels simultaneously.
To provide sufficiently high bandwidth road surface data in real time, his company decided that it could not rely on the industry standard, Open/CRG because it was too slow: “So we developed a solution using the proprietary platform, Terrain Server, that runs four times faster than CRG and is multi-core compatible; so with four cores, we are 16 times faster.”
An accurate road surface model is essential for the simulation of response to road inputs, such as ride and impact harshness. Because DIL applications have to run in real-time, delivering good quality feedback in the upper frequencies requires the rapid transfer of large volumes of data into the vehicle model.
“One of the most important external factors affecting the vehicle model is the road surface that the tires' contact patches traverse,” explains Hoyle. “Conventional single or multi-point contact models are adequate for low frequency applications but feel too harsh to a human driver. By integrating cleaned LIDAR point-cloud data from the road surface, captured via laser scanning of the road in real time with very accurate resolution of ‘z’ heights, we provide higher levels of data.”
The high bandwidth data feed to the vehicle model means that the limiting factor is now the tire model and not the road surface. The bandwidth available from rFactor Pro is described as being sufficient to support the modelling of NVH.
The overall response of a simulator is driven by both the software speed and the inertia of the motion platform itself. Older platforms, evolved from aerospace technology, may weigh up to 4 ton (3.6 t); however, the latest generation can weigh only 250 kg (550 lb).
The rFactor Pro system is so fast that it can keep ahead of even the fastest motion platforms, claims Hoyle: “In fact, we are helping to reduce platform inertia by mounting all our graphics off the platform. This is made possible by using optically correct curved images, presented on cylindrical or spherical screens.”
The company works in partnership with motion platform producers including McLaren, Moog, and Ansible Motion.
For potential business users of any new technology, questions invariably arise over its compatibility with existing investment; organizations may have many man-years invested in existing vehicle models and any simulator development that required a company to abandon fairly recent and often considerable previous investment, would face a serious obstacle. So rFactor Pro’s DIL simulation tool was designed to plug into and wrap around existing vehicle models, such as SIMPACK, Dymola, CarMaker, CarSim, Simulink, VSM and VI-Grade.
Hoyle feels that with the growth of electronic control across vehicles and the increasing complexity of CAE models, solving the physical simulation challenges will unlock a flood of new possibilities to time and cost reduction.