…the flexible open source driving simulation

The Continuous Tracking and Reaction Task (ConTRe Task) is a highly controlled driving task that allows fine-grained measurement of steering and event detection performance through very few dependent variables. Also, it is a rather artificial task that only resembles to realistic driving. NOTE: This driving task is only available in the professional version of OpenDS.

 

General idea and goals of the ConTRe Task

We developed a task that resembles following a curvy traffic way, with a lead vehicle braking and accelerating from time to time. Moreover, crossroads traffic or other peripheral events occur from time to time. Our goals were:

·       Creation of a highly controlled driving task;

·       Fine-grained measurement of steering and event detection performance (few dependent variables);

·       Adaptability of driving task difficulty and event rate (several levels, even on-line manipulation) to enable assessment of moderating effects of driving task difficulty (workload).

In cases where continuous access to steering performance or reaction times upon unforeseeable events are needed or whenever moderating effects of driving task difficulty on performance are in focus, the ConTRe Task is able to complement the de facto standard Lane Change Task and to provide more fine-grained insights into potential effects on performance.

Scenario

A driver's primary task when performing the ConTRe Task in a simulator is comprised of actions required for normal driving: The participant turns the steering wheel and operates brake pedal and accelerator pedal. System feedback, however, differs from normal driving. In the ConTRe Task, the car (where the camera is located) moves along a two-lane unidirectional straight roadway (Option C) at a constant speed. Neither operating the accelerator nor the brake pedal does have an effect on longitudinal control of the vehicle. Braking and using the gas pedal do not actually control the car, but instead these responses are given upon discrete events in the simulation. Turning the steering wheel influences the lateral position of the vehicle as well as the lateral position of a blue bar, which is located in front of the car and will always be in-line with the vehicle in terms of lateral position.

At the same longitudinal distance the driver perceives another yellow bar that is moving autonomously. The driver controls the lateral position of the blue bar via the steering wheel, trying to keep it overlapping with the reference bar center as well as possible. In the full version of the ConTRe task, there is furthermore a signal light placed on top of the reference bar showing one of two lights from time to time. The upper red light requires an immediate brake reaction with the brake pedal, whereas the lower green light indicates that an immediate gas pedal action is necessary (cf. Figure 2). Even though motion in the ConTRe task rather feels like a video clip, effectively, this task of continuous manual control based on continuous visual input and pedal responses upon discrete visual events roughly corresponds to a task where the user has to follow a curvy traffic way with a lead vehicle braking and accelerating from time to time. Even more important, it is absolutely controlled and leads to less user-dependent variability in interpretation of instructions. Furthermore, the task allows for (online) manipulation of task difficulty. Three predefined levels of task difficulty are provided (easy, medium, hard). 

Interaction concept

Turning the steering wheel causes normal lateral movement of the car. In addition, the blue bar moves accordingly. The instruction for the participant is: “Keep blue bar in the middle of the yellow reference bar by steering”. Occasionally, the traffic light turns either red or green. When the red traffic light appears, the participant has to press the brake pedal. When the green traffic light appears, pressing the gas pedal is required. Also, occasionally, the peripheral icons appear either red or green colored. Equivalently, the driver has to press the according pedals as mentioned above.

At the same longitudinal distance the driver perceives another yellow bar that is moving autonomously. The driver controls the lateral position of the blue bar via the steering wheel, trying to keep it overlapping with the reference bar center as well as possible.

Figure 1. Screenshot of the ConTRe Task including the central traffic light and peripheral events – driver’s view, the red light indicates that a brake response is required.

From time to time, either a red or a green automotive icon is displayed on the right or left patch. The red icons require an immediate brake reaction with the brake pedal, whereas the green icons indicate that an immediate gas pedal action is necessary (cf. Figure 1). These pedal responses upon discrete visual events roughly correspond to a task where the user has to suddenly classify events in the periphery (critical or not critical) and conduct a respective response.

Figure 2. Screenshot of the ConTRe Task including central traffic light and peripheral events – driver’s view, the green icon on the left black square indicates that a gas pedal response is required.

Even though motion in the ConTRe task rather feels like a video clip, effectively, this task of continuous manual control based on continuous visual input and pedal responses upon discrete visual events roughly corresponds to a task where the user has to follow a curvy traffic way with a lead vehicle braking and accelerating from time to time. Moreover participants have to detect suddenly appearing events and they have to react adequately. Even more important, it is absolutely controlled and leads to less user dependent variability in interpretation of instructions. Three predefined levels of task difficulty are provided (easy, medium, hard). We chose the medium level for our experiment since this level is the most suitable for a dual task situation. While performing the task, mean deviation from perfect bar overlap (blue bar covers both black dashed lines on yellow bar) and discrete brake and gas pedal reaction times (start-point: target event onset – first point of time when information on dedicated reaction is presented; end-point: First brake/accelerator pedal contact) are recorded in a database. Table 1 illustrates parameters of the medium ConTRe Task level for a drive with a duration of 3 min.

Dependent Variables

Generally, there are two distinct dependent variables:

Firstly, mean deviation from perfect bar overlap (blue bar covers both black dashed lines on yellow bar). The average value can be calculated for a complete track or a defined period within a track.

Secondly, reaction times for discrete brake and gas pedal usage can be analyzed. The start point is defined to the target event onset (first point of time when information on dedicated reaction is presented) while the end point corresponds to the first brake or accelerator pedal contact.

While performing the task, mean deviation from perfect bar overlap (blue bar covers both black dashed lines on yellow bar) and discrete brake and gas pedal reaction times (start-point: target event onset – first point of time when information on dedicated reaction is presented; end-point: First brake/accelerator pedal contact) are recorded in a database. The next section presents the recorded variables in more detail.

With the ConTRe Task, several dependent variables can be assessed. First of all, the mean steering deviation from perfect bar overlap (blue bar covers both black dashed lines on yellow bar) can be assessed either averaged for a complete track, or averaged for a defined period within track. Secondly, reaction times are recorded (separately) for central and peripheral events via discrete brake and gas pedal reactions. Here, the start-point is always the target event onset (i.e., the first point of time when information on dedicated reaction is presented), and the first contact with the correct brake or accelerator pedal denotes the end-point of time measurement.

Every next designated location of the yellow reference bar and thus its movement direction are neither controlled nor predictable by the user. The only exception is that lateral movement of both bars is limited by the solid side markings of the travelled way, which are eight meters away from each other. This prevents the driver from leaving the road (unintentionally). The driver controls the lateral position of the blue bar via the steering wheel, and tries to keep it overlapping with the reference bar center as well as possible. As the maximum speed of the steering bar is 4.0 m/s and thus 10 times higher than that of the reference bar, 1/10 of the maximum steering wheel displacement is sufficient to keep the steering bar follow the reference bar at the same speed. In the full version of the ConTRe task, there is furthermore a signal light placed on top of the reference bar showing one of two lights from time to time. The upper red light requires an immediate brake reaction with the brake pedal, whereas the lower green light indicates that an immediate gas pedal action is necessary (cf. Figure 2). In order to additionally test peripheral vision we used a ConTRe version with a third event category in our experiment. In this extended version of the ConTRe Task participants do not only have to use the brake and gas pedal when the central traffic light turns red or green, but also when a peripheral symbol appears at the edge of the field of vision. A black area on the right and left side of the screen indicates where the symbols will appear. The symbols are green or red and show different objects known from road traffic. Participants are instructed to use the brake pedal when a red symbol appears and to use the gas pedal in case of detecting a green symbol.

Technical instructions and parameter settings regarding the ConTRe task

 

Parameter

medium level

Speed of reference bar (unit: m/s)

0.4

Time of rest till reference bar moves to next (random) position (unit: s)

2.0

Time of intervals between events (unit: s)

min: 4

max: 20

Number of events overall

7 - 19

Number of peripheral events

3 - 10

Number of central events

3 - 10

Duration of an event if no reaction is recorded (unit: s)

5

 

Initial assessment of the driving task

ConTRe was designed to be a rather artificial task that only resembles to realistic driving. The goal was to create an absolutely controlled task and clear task description to generate less user-dependent variability in interpretation of instructions. Three predefined levels of task difficulty are provided (easy, medium, hard) to serve for diverse application purposes.