Ford Motor Company Research & Advanced Engineering Physical Fidelity and Driving Simulation 21 things to know about your simulator before you write the

Ford Motor CompanyResearch & Advanced Engineering

Physical Fidelity and Driving Simulation

21 things to know about your simulator before you write the

proposal

Jeff Greenberg


fidelity n 1: accuracy with which an electronic system reproduces the sound or image of its input signal 2: the quality of being faithful [syn: faithfulness] [ant: infidelity,infidelity] Source: WordNet ® 1.6, © 1997 Princeton University


How much is enough?

● Depends on the experimental details, of course!

● Simulator experiments generally seek to measure a human response across a variety of controlled conditions

● The goal is often to extrapolate those findings from the simulator to the traffic system at large: real people, real cars, real roads.


Fidelity Spectrum

● Real People, Real Cars, Real Roads

Epidemiology (Truth Yesterday)

● Sampled People, Real Cars, Real Roads

Black box observation, naturalistic data collection (Truth Today?)

● Sampled People, Fake Cars, Real Roads

Prototype studies, Field Studies (Truth Tomorrow??)

● Sampled People, Virtual Cars, Virtual Roads

Driving Simulation (Truth Today, Tomorrow and Beyond ???)

FID

EL

ITY

higher

lower


Infidelity or Artifacts are key

● All simulators have artifacts

● The goal of characterizing a simulator’s fidelity is to know what your major artifacts are and how big they are in the context of a given experimental setup.

● The key to experimental success is understanding how those artifacts interact with your experimental design


The Art of Simulator Triage

No matter how compelling the research interest, a simulator study should only be done if you understand the artifacts that are likely to occur in your simulator during the experiment and have the ability to control them to an acceptable level.

In practice, most patients cannot be saved – i.e. most research questions cannot be answered in a given simulator with a useful level of accuracy.


How do I know what artifacts to expect?

● Start with a knowledge of the most important characteristics of physical fidelity in your simulator

● Cueing systems

Visual

Auditory

Haptic or Tactile

Vestibular

● Physical environment

Cab

Laboratory

● Virtual environment

Visual world

Terrain and roadways

Vehicle Models


Visual SystemCharacteristic Key Measure Most Likely Artifacts

1 Horiz. Field of View Angular extent Too Small: not enough visual information to see or drive scenario. Improper mirror scanning behavior and visual allocation.

Too Large: potential simulator sickness

2 Perceived resolution Angular extent of the smallest resolvable optical line pair

Too Small: Improper threat identification. Difficulty seeing roadway and signage. Eyestrain & fatigue, simulator sickness.

3 Transport delay Time delay between operator action and visual system response

Too large: Degraded vehicle control. Increased simulator sickness.

4 Accommodation Distance at which eye accommodates

Too Small: eyestrain, poor depth perception and distance judgements, simulator sickness?


Auditory SystemCharacteristic Key Measure Most Likely Artifacts

5 Wind/Tire noise loudness Loudness as a function of speed

Too Small: Underestimation of vehicle speed

Too Large: Discomfort

6 Powertrain loudness Loudness as a function of engine speed and load

Too Small: Poor onset acceleration cue


Haptic SystemsCharacteristic Key Measure Most Likely Artifacts

7 Steering loader torque Torque as a function of vehicle speed and sw angle

Too Small: poor vehicle control

Too Large: difficulty steering, potential safety concern

8 Steering loader damping Free response to small step input

Too Small: instability, potential safety issue

Too Large: steering feels ‘dead’, poor steering control during lane changes

9 Brake pedal stiffness Force per unit distance

Too Small: low initial braking forces

Too Large: high initial braking forces, can’t modulate brake pedal, increased simulator sickness


Vestibular CueingCharacteristic Key Measure Most Likely Artifacts

10 Useful motion range (<3Hz) Maximum useful acceleration scale factor per degree of freedom

Too Small: poor vehicle control, simulator sickness, visual demand too high

11 Small Signal bandwidth 3dB frequency

90deg phase

Too Small: incorrect ride dynamics, poor road feel, ‘magic carpet’ ride

12 Roughness Acceleration noise (g’s)

Too Large: false cues, poor on-center control, discomfort, anxiety


Environment: Cab & Laboratory

Characteristic Key Measure Most Likely Artifacts

13 Realistic greenhouse A-pillar geometry No A-pillars: incorrect FOV, changed visual scanning

14 Visual Immersion Number of objects fixed in laboratory frame that are visible while driving

Too Large: destroys perception of self-motion in fixed sim, in moving sim generates false motion cues


Visual World


15 Maximum traffic density Number of cars in scene without frame overruns

Too Low: may not meet experimental objectives. Adding traffic and tolerating overruns may invalidate vehicle control measures and increase simulator sickness!

16 Clarity of road markings Distance at which road markings are clearly visible

Too Small: Drivers may be unable to determine correct vehicle path


Terrain & Roadways


17 Offroad capability Can driving occur anywhere in the world or only on ‘roads’?

‘Road’ only: agricultural and off-road simulations not possible

18 Terrain fidelity Spatial frequency of the terrain geometry

Too Small: Vehicle motions unusually rough over terrain that appears to be smooth. When coupled with vibration or motion simulation can induce discomfort.


Vehicle Models


19 Acceleration 0-60 time (s) Too High or too low: Unrealistic following abilities and gap acceptance.

20 Engine braking/Aero drag Decleration at closed throttle as a function of vehicle speed

Too High: Unrealistic car following

Too Low: Unrealistic car following, overuse of brakes to maintain speed

21 Lateral acceleration gain Ay/Swa as a function of vehicle speed

Too Low: Incorrect lane keeping, poor lane change ability

Too High: Incorrect lane keep, unrealistic lane change ability, potentially unstable

Documents

Ford Motor Company Research & Advanced Engineering Physical Fidelity and Driving Simulation 21 things to know about your simulator before you write the