The Exploratory Advanced Research Program

Making Driving Simulators More Useful for Behavioral Research

Simulator Characteristics Comparison and Model-Based Transformation

PHYSICAL FIDELITY

Physical fidelity relates to the degree to which the simulator replicates the physical properties of the driving situation, unlike behavioral fidelity, which is associated with the simulators’ ability to replicate behavior observed in the world. This study’s research team examined four simulators representing a broad range of simulation capability and fidelity and measured characteristics for each. Simulators included in the study were the National Advanced Driving Simulator (NADS), the FHWA Highway Driving Simulator, the Western Transportation Institute (WTI) Simulator, and the NADS miniSim.

Study Simulators

The following section provides a brief overview of each of the four simulators used in this study.

National Advanced Driving Simulator

The NADS used a 1998 Chevrolet Malibu cab mounted on a motion base with 13 degrees of freedom, as shown in figure 1. Accelerator and brake pedals used software-controlled electrical motors to provide feedback. The simulator has a 360-degree visual display system consisting of eight projectors that project visual imagery inside the dome, and scenery is updated at 60 Hz. The NADS features the ability to swap among several types of vehicle cabs.

Figure 1. The National Advanced Driving Simulator motion-base driving simulator.

Federal Highway Administration Highway Driving Simulator

The FHWA Highway Driving Simulator, shown in figure 2, is composed of a full 1998 Saturn vehicle cab mounted on a motion base with 3 degrees of freedom. The FHWA simulator has a 240-degree visual display system consisting of five projectors that project onto a cylindrical screen that is 2.7 m (9 ft) tall. All scenery is updated at 60 Hz.

Figure 2. The Federal Highway Administration motion-base driving simulator.

Western Transportation Institute Simulator

The WTI simulator, shown in figure 3, consisted of a 2009 Chevrolet Impala sedan mounted on a motion platform with 6 degrees of freedom. The WTI simulator has a 240-degree forward-field-of-view system augmented by a 60-degree rear-view display system, consisting of five projectors and a curved screen in front of the driver and a single projector and a flat screen behind the driver. Side-view mirrors with digital screens also portrayed the scenarios for a total of eight visual channels.

Figure 3. The Western Transportation Institute Simulator.

National Advanced Driving Simulator miniSim Simulator

The NADS miniSim, shown in figure 4, is a portable, lower cost simulator that runs software similar to the NADS simulator. The miniSim has no motion base and, for this study, featured a quarter-cab configuration with a seat and steering wheel from an actual vehicle. It has three flat-panel plasma displays and projects the image of a rear-view mirror on the center plasma display.

Figure 4. The National Advanced Driving Simulator miniSim Simulator.

Supporting a Realistic Driving Simulator

Simulators are often characterized by a set of features that describe their hardware components, including driving controls, screens, resolution, and mirrors. The hardware configuration is critical for conveying information to the driver, such as speed and curve geometry and gas and brake pedal force; however, although hardware is a necessary condition for high behavioral fidelity, it is not sufficient on its own. For a driving simulator to accurately convey the driving environment, it must depend on hardware and software. In fact, software is often more important than is hardware because it is the software controlling what is presented to the driver.

Working together, the hardware and software generate signals for the driver and influence how they perceive the environment and control the state of the vehicle relative to the environment. There are three key requirements for supporting a realistic driving simulator: (1) perception of distances, speed, and time to reach relevant objects in the real world; (2) control of the car’s speed and direction through control inputs; and (3) vehicle response to the control inputs.^12,16,17 The ability to identify and measure simulator characteristics in relation to these requirements enables researchers to define important differences between simulators, even if their hardware specifications are identical. In addition to taking a sample of measurements to quantify the physical fidelity of the driving simulators used in this study, the research team aimed to relate each simulator characteristic to what drivers experience on the road to assess how characteristics might affect behavioral fidelity.

Measuring Levels of Realism

Following data collection and assessment of simulator characteristics, the NADS and WTI simulators showed the highest level of physical fidelity; however, study results indicated that no single metric can serve as a proxy for overall simulator fidelity. In fact, the broad concept of overall level of fidelity is in fact misleading and should instead be addressed in a multidimensional manner. Several issues must be addressed before this multidimensional approach is applied more broadly. For example, cars differ substantially across most of the simulator characteristics measured. It is therefore important to identify which differences are important and which are not. Drivers were also shown to easily adapt to a wide range of vehicle characteristics, including maximum acceleration and deceleration, steering inputs, pedal feel, and visual contrast. Even though a driver might perceive differences between a simulator and the car on the road, the difference might not influence driver behavior but may still result in differences in workload and driver strategies in obtaining the same driving performance. Following analysis of various metrics of physical fidelity, further research is needed to quantify the variation of simulator characteristics, the degree to which drivers can adapt to different vehicle simulator characteristics, and the degree to which these characteristics influence behavior, driving strategies, and operator workload.