Driving Dynamics Lab Facility

Driving Simulator

At the center of the Driving Dynamics Lab is the driving simulator. Our driving simulator uses high-powered computers to allow high fidelity vehicle models to interact with the virtual world. These vehicle models are highly tuned and use industry-standard programs that separate this simulator from an entertainment simulator.  Accurate dynamics provide realistic driving conditions. Whether accelerating quickly around a turn or bumping a curb, the driving simulator captures all these responses and sends them to the driver for feedback. Emphasis has been placed on realistic models allowing for dynamic loading of the vehicle.

Typical human-machine interfaces such as the instrument panel cluster, heads-up display, steering wheel buttons and audible warnings are used for immersion. Redundancy of warnings for the severity of system messages is toggled to alert the driver to direct behavior. As with all simulator work, there are limitations to what a simulator can accomplish. Work with a driving simulator needs to ensure tuning of a model rather than tuning of a simulation. Detailed tracking of the assumptions used to create the DiL’s physical and virtual features alleviates this issue. This DiL was designed to give the driver the sense of being in a real vehicle but done so with a realistic approach to the intended uses.

Researchers use both fictitious and real virtual terrains that a driver can drive. Real virtual terrains are models of real world locations. These terrains cover the typical driving environments that are crucial to vehicle dynamics development along with autonomous test tracks. Scenarios are created with these terrains to aid developmental work. Specific use scenarios are created to induce the intended response by the system and driver. Further, the repeatability of these test scenarios is guaranteed because there is complete control of the simulated environment. For example, surrounding autonomous vehicle and pedestrian traffic is used to evaluate the ability of the different systems. These tools can be combined with changes in weather, light, road surface definitions, and intersection types to name a few. Combining these variables allows for controlled, repeated tests.

Further, controller design is streamlined with a DiL. Controllers regulating the vehicle can be designed to accommodate a few of the subjective metrics that most passengers require. For instance, quick changes in acceleration have been linked to an uncomfortable ride, so a DiL is invaluable for evaluation of when changes in acceleration become too severe. Similarly, other interactions with the ADAS system of interest can be observed. If a driver wants an override control, then that override control can be identified and incorporated earlier in the design. Contrarily, if the control needs to be added to override a driver, then such control can also be incorporated earlier in the design.

Cueing Systems

Visual System

The most important cue for the driver is provided by the visual system. As in actual vehicles, path control and sense of path error is primarily visual.  As with all of the cueing systems, particular attention has been placed on reducing latency to the driver.  The VDDiL displays are low latency and the graphics system is updating them at 120 Hz.

Torque Feedback Steering Wheel

Another important cue is the steering wheel. Drivers are sensitive to the steering response. The motor connected to the steering column allows for precise control of the steering wheel's torque feedback. 

Motion Platform (3 - degrees-of-freedom)

Vehicle response of the model is sent to the platform to change heave, roll, and pitch. These changes are intended to signal to the driver changes that happen to the vehicle model. Changes in acceleration and velocity are not delivered in the computed magnitude. Rather, vibrations and quick maneuvers interact with the driver's vestibular system.

Seatbelt Tensioner

Simulating how a driver is restrained by the seatbelt in a braking maneuver, the seatbelt tensioners pull the belt into the driver to simulate longitudinal and vertical acceleration cues. This minimal tensioning gives a sense of the vehicle's acceleration.

Human-Machine Interface

Instruments from the Ford Mustang buck have been integrated with the simulator. This integration allows for use of the dash components, heads up display, steering wheel buttons, and more. Matching these displays in real-time helps align the virtual driving experience to its real world counterpart. 

Simulator Configurations

Currently, the simulator uses 2 distinct software configurations to render the virtual world. Both configurations are built on the same vehicle modeling software but are used to research different areas. Testing and developing advanced driver assistance systems is done with SCANeR Studio from AV Simulation. SCANeR's developed sensor models resemble sensor models placed on real vehicles. The typical sensors that OEM's use to implement these advanced driver assistant systems can first be tested on the simulator in a controlled scenario and terrain. The code can be uploaded to check the response of different sensor measurements to aid developmental and validation work.

The other configuration uses a product made by Mechanical Simulation that pairs a physics engine with a gaming rendering software. This rendering software generates the image that the driver sees of the virtual world but matches what the model querries. This modular configuration allows the team to change important behavior. Combining the two configurations enable the team to be dynamic in their testing procedures. It is also useful to have the ability to compare vehicle responses between the two configurations for a source of validation.