The emergence of Advanced Driver Assistance Systems (ADAS) and Automated Driving (AD) systems is gradually preparing consumers for a time where they relinquish control of their vehicles. A heated source of debate has been whether drivers are ready to hand power over to autonomous vehicles. But what about the other way around? What are the technology implications for an Autonomous Vehicle (AV) handing control back to the driver in a perceived emergency situation?

By providing precise, on-time cueing (audio, visual, haptic, vestibular), the aVDS immerses the driver in the experience, creating the impression of driving. A high excursion capability ensures plenty of travel for effective motion cueing, enhancing the impression of changes in speed and direction. Low latency ensures that all the driver's senses are cued at precisely the right time.

“In a level 3 and 4 automated vehicle, it is easy for the driver to lose perspective of speed, road positioning, and their immediate environment when they are reading a news article, checking emails, or engrossed in conversation. It takes time to reacclimatize,” said Professor Pim van der Jagt, Technical Director at AB Dynamics Europe GmbH.

The automotive test system supplier has recently provided one of its Vehicle Driving Simulators (aVDS) to Kempten University of Applied Sciences to help address these challenges. “If the system controlling the vehicle is unexpectedly unable to cope with an impending situation, it alerts the driver to the need for them to take over, but it takes the driver time to assess and fully understand the situation,” explained van der Jagt.

“How fast is the vehicle traveling? How are the surrounding vehicles behaving? Is the vehicle approaching a junction or traffic lights or bearing down on a vulnerable road user and is the driver in a position that is optimal for regaining control?” van der Jagt said. “For both the vehicle's systems and the human brain, there is vast information to process and potentially an incredibly small amount of time to do this.”

The primary challenge, according to van der Jagt, is how to assess the optimal method for capturing the driver's attention, making them quickly aware of the scenario, and finally, handing back control.

Simulation is Key

A car-like cockpit module is mounted on a motion platform that is controlled by four ‘wedge’ actuator modules. This assists with immersion, which is critical to maintain the authenticity of the driver's experience and reactions. Quiet and lightweight linear motors control both the height of the platform on the wedge and the position of the wedge on the rail.

The holistic human-machine interface (HMI) concept — consisting of an ideal mix of acoustic, visual, and haptic alerts — is very important. In order to redirect the occupant to the driving task required, it is essential to understand the occupant's behavior in these situations and this must be intensively considered within a human-centric approach. There are differing ways in which the need for control handover can be signalled; for example, it could be via vibration or movement of the occupant's seat or via an audible or visual alert, or a combination of these.

These high-risk scenarios cannot be replicated on the public road, which leaves prototype and test center validation. Or does it? These complex and diverse scenarios are also very difficult to replicate on proving grounds.

“Maneuvers must be undertaken on a proving ground using prototypes fitted with automated systems and functions, which is a costly and time-consuming option,” said Professor Bernhard Schick, University of Applied Sciences Kempten. “Prototypes are therefore becoming less favored for this type of testing, but it remains vital for engineers to experience and evaluate new functions in situ. Therefore, third-generation driving simulators provide an increasingly attractive solution for researchers at centers such as our Adrive Living Lab. This facility, which is nearing completion, will include AB Dynamics’ advanced Vehicle Driving Simulator, and will focus on the development of autonomous driving,” Schick explained.

“A key benefit of using a driving simulator is to explore the subjective as well as the objective effects of driver inputs under a diverse range of conditions and circumstances. The motion platform of the aVDS,” Schick said, “offers excellent excursion, which complements its high dynamic capability and extremely low latency. This means driving inputs are transferred into vehicle motion with minimal time delays or cumbersome-feeling dynamics; the resulting experience is precise and ultra-realistic.

At the core of the simulator is the motion platform. The dynamic performance and working envelope of the platform are optimized for vehicle driving simulation, leading to a powerful yet compact solution. Its unique kinematic mechanism ensures a consistent and linear response throughout the range of travel, providing accuracy no matter where the platform is.

“When driving on the aVDS, the driver is fully immersed and provided with an environment that can be as much like traveling within an AV as possible,” Schick explained. “This immersion is key, as it means the driver will react completely naturally to the test situation, just like they would in a real vehicle. The dynamic ability of the simulator — combined with its visual, audio, haptic, and vestibular cueing systems — provides drivers with a level of authenticity that evokes genuine responses and reactions to simulated scenarios. If the AV hands back control in a range of conditions, operators can assess the series of events and eventual impact on the driver, passengers, vehicle, and surroundings.”

How it Works

Historically, simulator platforms were based on the hexapod architecture that originated in the aircraft industry, where the mass is higher and rates of turn much slower. While the rate of response of these simulator platforms was sufficient for automotive vehicle dynamics applications, the horizontal excursion capability required for driving simulation was too limited. This has led to the development of a third generation of compact driver-in-the-loop simulators, such as the aVDS, which have both good excursion capability as well as excellent dynamic performance.

The vehicle cockpit of the aVDS is mounted on a motion platform positioned by wedge actuator modules that run on two parallel rails synchronized by the motion control system. The modules control both the height of the platform on the wedge and the position of the wedge on the rail. The platform has angled sides, so it can be moved backwards and forwards by changing the distance between the wedges on each rail. Turning the platform is achieved by moving the front and rear actuator pairs in opposite directions along the rails. Heave, pitch, and roll are controlled by moving the wedges independently to lift or lower the platform at each corner. This design is ideal for a vehicle driving simulator because the motion ratio is very linear throughout, meaning the dynamics performance (up to 60 Hz bandwidth) is consistent throughout the full range of travel of the motion platform.

Motion cueing algorithms are used on a driving simulator in order to provide motion cues that make the simulator feel as much like a real vehicle as possible, while remaining within the physical limits of the platform and the limitations of the vehicle model. The vehicle model is used to predict the response of the vehicle to the driver inputs. The cueing algorithm then converts this vehicle motion into the motion of the simulator platform required to give the desired driver sensation.

The aVDS software, supplied by rFpro, provides a rich and detailed world, with data taken from exhaustive analysis and measurement of terrain. The road surface information has been collected from standard test tracks, proving grounds, and even real road networks and can be presented to the driver with extreme accuracy.

The software in the aVDS, provided by rFpro, delivers video and audio signals significantly faster than other systems. By focusing on speed of response and video bandwidth with minimal latency, rFpro allows the aVDS to be used in highly dynamic situations, enabling simulation of anything that affects the dynamic behavior of the vehicle.

The software also incorporates the output from a 3D mapping process employing LiDAR laser imaging technology to capture physical road and track surface data in sufficient detail to run complete contact patch models for all four tires in real time on an actual road surface. Every feature of the surface is reproduced in 3D with an accuracy better than 1 mm in Z at 5-mm intervals across the road surface.

The simulator's modular architecture can also be utilized for additional benefits. Kempten is installing a physical steering system from wheel to ball joints to further enhance feedback and data capture. In the case of AV evaluation and development, the hardware-in-the-loop capability of the aVDS is particularly valuable for difficult-to-model, non-linear components, as it improves the realism of the simulation as well as driver immersion and the accuracy of the results.

Conclusion

Utilizing driving simulators provides the perfect opportunity to replicate the myriad of ‘handing back control’ scenarios that may occur, both in terms of repeatability and accuracy; however, it also offers the additional benefits of being able to negate the need for expensive prototypes during the development phase, and can ultimately accelerate the time to market for next-generation autonomous vehicles via a streamlined methodology for overcoming the challenges of emerging technology. If the role of autonomous vehicles is to make our roads safer, driving simulators could well be the enabling technology that delivers these benefits in a safe, affordable, and timely way.

This article was contributed by AB Dynamics Europe GmbH. For more information, visit here .