Engineers who work with driver-in-the-loop (DIL) simulators all have opinions on the different motion systems and motion-cueing models that are available on the market. Some of these viewpoints are based on experience, but supposition, rumor and hearsay can also be factors.
Cruden has been working on motion simulators for around 25 years, having installed more than 100. Our team of 25+ people is dedicated to improving the realism and accuracy of vehicle simulation. The focus is on motion that makes the driver behave realistically. Different driving situations have different driver behaviors; different motion-cueing strategies are therefore required for performance and urban driving in research, vehicle-engineering and driver-training applications.
Cruden is known for building simulators around the six-degrees-of-freedom (6-DOF) hexapod motion system. At the recent AIAA SciTech 2019 Forum, Cruden set out its motion-cueing philosophy in a paper entitled, Adaptation of the Classic Cueing Algorithm for Automotive Applications. This article summarizes the approach that Cruden has taken, and why.
What’s the problem?
Drivers feel the motion of a vehicle and respond to it. Performance drivers do that with great attention to detail, while commuters respond without even realizing. The behavior of every driver is affected by what he or she feels, both in a car and in a simulator. That’s a major challenge, because the goal in every DIL simulation is to make a human driver behave as naturally as they would in a real vehicle.
Simulators must convert a real-world workspace into a motion-system workspace, which in turn necessitates a mathematical solution to convert vehicle motion into platform motion. This is known as motion-cueing. Often, these cueing algorithms induce discrepancies between individual DOFs, provide false cues, or create other issues.
Driver behavior is influenced by these unwanted factors. The best-known influence is simulator sickness, but unnatural driver behavior starts to occur long before the driver feels unwell. In fact, driver sickness is caused by the simulator as a whole. Visuals that are not synchronized with the motion can just as easily cause sickness as false motion cues.
Many simulators, with a variety of hardware configurations, utilise what is commonly known as the Classic Cueing Model. This is a mathematical model in which vehicle acceleration is manipulated using filters and gains to create platform motion.
There are issues with the Classic Cueing Model, which is derived from aerospace use. Some simulator engineers overcome this by implementing a Model Predictive Control algorithm (MPC), which optimizes platform motion to its own prediction of the future movements of the virtual vehicle, in real time. With MPC the focus is on scaling the acceleration to fit the available workspace of a simulator. The downside of this algorithm is that every corner will feel the same, which makes the motion less relevant.
The solution: Cruden Cueing Model
At Cruden, the aim is always to achieve the best possible driving experience, not to replicate the vehicle as accurately as possible, at any cost. These two concepts are closely related, but not always the same. As a result, the approach to motion cueing is subjective, relying heavily on driver behavior and feedback.
By adjusting the Classic Cueing Model, DOF by DOF, improvements are possible. Here’s what Cruden has done.
Pitch and roll
The pitch and roll cues in the classical cueing model are created through the filtering of pitch and roll accelerations. These washout filters help the platform stay within its limited range of motion, for example by slowly and automatically returning it to a neutral attitude during a prolonged turn. However, this can lead to situations where the platform movement is in the opposite direction to the vehicle movement – a false cue.
Cruden’s approach is instead to cue pitch and roll based on a position signal. In tests, simulator drivers all preferred this position-based cueing.
Racing drivers often indicate that a good braking cue gives them “confidence on the brakes”. If the driver is to behave as naturally as possible in the simulator, a good braking cue is essential in performance-driving scenarios.
A typical issue with simulators is that the platform stops and returns to centre while the driver is still braking. Cruden has optimized the workspace management system so that the platform will move to provide the braking cue and then come to a stop near the end of the workspace, but not move back towards the centre. After braking, an additional, artificial driving signal is added to the cueing output while the vehicle is accelerating, to pre-position the platform for the next input. Steps have also been taken to remove false cues that occurred after a heavy braking phase. With Cruden Cueing, the platform will only move forward when the driver is accelerating, not when they release the brakes,
Driver feedback on this new combination of surge cues has been very positive, with inexperienced drivers noting that braking and acceleration feels natural, and performance drivers reporting that they have confidence on the brakes.
A vehicle’s typical pitch and roll angle behavior fits well within the motion system’s workspace, but the vehicle yaw angle – or heading angle – does not. Cueing the heading angle directly is therefore not an option in most driving simulators, but Cruden has explored several methods to provide cueing solutions that feel natural to the driver.
Engineers established that the traditional approach of filtering the yaw acceleration resulted in false cues due to washout and necessitated overly conservative tuning. This is a particular issue for performance driving. Here, Cruden instead uses the body sideslip angle for yaw cueing. This provides the driver with unfiltered information on the vehicle performance limit but omits heading changes.
For everyday road driving, an alternative cueing method is implemented, with the heading angle being cued directly when driving on a straight highway, or cued from a processed combination of heading, yaw-rate and yaw acceleration signals in low-speed turns. The result of this heading-based approach is that the direction of platform rotation more closely matches the direction of the vehicle rotation throughout the maneuver, which makes the platform’s behavior feel more natural to the average commuter.
Using the classical algorithm, the platform’s swaying motion suffers from onset cues that are inconsistent with vehicle behavior, specifically when the driver turns the steering wheel to drive straight ahead after a long turn. Cruden engineers have applied a remedy, using a body sideslip measurement to modify the sway cue specifically for this case.
Similar to the yaw-cueing approach described above, further modifications tackle the highway and performance-driving use cases. In testing, simulator drivers recognised that the synchronization between the yaw and sway cues had improved, which made the front end of the vehicle feel more responsive. As in a real car, the yaw and sway motions are now cued in a synchronized way and therefore feel natural, whereas before, the two cues appeared almost completely unrelated.
In the motion cueing evaluations conducted with Cruden simulators, the heave motion from the classical algorithm was not identified as an area with problematic behavior, so it has not been adjusted or replaced. However, the heave channel’s provision of road-noise information to the simulator driver is an important cue for vehicle velocity that is often missing from simulations, so Cruden has enhanced it by adding a road-noise generator to the cueing model.
Cruden has adapted and augmented the classical cueing model in all degrees of freedom to make it more suitable for vehicle simulation. For each DOF, the underlying philosophy is the same: to ensure the direction of movement of the platform no longer opposes the direction of movement of the simulated vehicle; and to remove differences in phase between individual degrees of freedom. The result is that drivers behave more naturally behind the wheel of the simulator.
This cueing model has been developed with and evaluated by 15 professional racing drivers, as well as by a similar number of inexperienced driver subjects. One professional driver said, “The road-noise, the pronounced kerbs, the aggressive movements, all make it harder to place the car exactly where you want it – which is what happens in the real car.”
The Cruden 6-DOF hexapod motion platforms have proven to be very effective in automotive driving simulators and are the basis of many of the company’s installations over the past 20 years. However, the Cruden approach to motion cueing works on any type of motion platform and has been implemented to enhance the usability of various existing simulators. Besides, Cruden has engineered custom motion configurations for specific use cases like ride and handling research.