Directly comparing a sim to the real world could open up invaluable opportunities for simulator refinement
During my career, I have spent a considerable amount of time operating a vehicle in a hexapod-supported driving simulator with extensive lateral motion capability. The performance of this particular simulator was astounding. The 360° visuals with no perceived latency and the operative vehicle inside the pod with a fully functioning dash came close to reproducing a true experience.
One improvement made during this time was an audio simulation refinement that enabled the noise signature of each tire to be individually simulated from the proper quadrant within the car instead of giving general tire noise feedback. As a dynamicist, listening to each tire ‘talking’ is one of the observations I rely on as a dynamic maneuver progresses.
The simulator’s travel, velocities and maximum acceleration capabilities over time are limited compared with the real world, so the dynamic performance and dynamics sensations typically need to be scaled down depending on the maneuver. Driving on a winding road requires a greater scale-down in capability than driving on a highway. It is even possible to perform steady-state handling maneuvers by biasing to one side of the road and driving in an arc across the two lanes.
In my experience, driving simulators are useful in at least five fundamentally different areas:
They enable research of human behavior and human-vehicle-situation interactions in scenarios that would be dangerous or cumbersome to undertake in the real world.
They can be used to familiarize operators with cockpit controls and vehicle operational procedures, including internal and external distractions and disturbances.
They can be used to train drivers, supporting them in memorizing track layouts and sight lines, and enabling them to develop their sequencing of steering, shifting, power and brakes, and cockpit operations at race speeds.
They provide a platform for evaluating vehicle dynamics architectures, control strategies and vehicle design variations in an easily reconfigurable environment.
Finally, they offer a training tool for developing vehicle dynamicists early in their professional careers.
Having been deeply involved in the latter two applications of a driver-in-the-loop simulator, I can share some experiences.
Evaluating various vehicle architectures, control strategies and design variations can be done on a sophisticated simulator as long as the scaling and latency do not mask the differences between iterations. It is amazing how perceptive professional vehicle dynamicists are. Performance differences that occur in less than ¹⁄₁₀ of a second and magnitudes of less than 5% can be obvious to a professional vehicle dynamicist who is intimately familiar with a particular vehicle. Users can also sense these differences in performance but cannot adequately describe their sensations. Regardless of how great the sim is, ultimately a real vehicle, in the real world, is a requisite for finalizing driver feel.
I have conducted training sessions for vehicle dynamicists early in their careers, before they had the opportunity to develop their repertoire of experience by building multiple vehicle platforms. We used a single vehicle model and changed parameters such as suspension compliance, kinematics, damping, and tire and power-steering characteristics to give them physical experience of how each parameter changes vehicle performance and sensations. We worked to build their repertoire of driving experience, which comes with time, tenure and opportunity, by providing similar experiences in the driving simulator.
One type of investigation I was not able to accomplish using the driver simulator was to instrument the simulated vehicle identically to the real vehicle and operate both with robotic steering in identical maneuvers. Here, the goal was to compare data traces directly between the driving simulator and the actual vehicle.
During my time in the simulator, the folks operating it were not focused on finding out how
it truly compares to the real world. Staying ‘blind’ was better. However, I bet that such a review would reveal significant observations, questions and opportunities for improvement to the simulator itself. I will leave this challenge to the simulator community.
