Keywords

1 Introduction

The change in the operation function of the occupant in the police riot vehicle (PRV) is the result of exploitation and evaluation of human’s ability in task. Occupant’s function has great changes compared with the initial design of PRV. Occupant’s function varies from operation to information processing and decision-making. In addition, coordination between vehicles becomes more and more important in the future, which may leads to decrease of occupant’s number and tightly integration of information interface. However, people’s ability especially information processing and analyzing has a certain limit [1]. Vehicle occupants also need to pay attention to changes in vehicle information during driving, and deal with a large number of unpredictable complex information caused by external environment quickly. Overload situation appears once occupant is assigned tasks which beyond his or her ability. Overload situation leads to errors, thus affecting the safety of occupants and vehicles.

Therefore, occupant’s function should be assigned according to occupant’s ability and workload impact factors, and determine optimal function allocation based on workload evaluation results. Occupant is freed from the complicated routine operation, and individual ability focus on information monitoring, collection, analysis and decision making. So, vehicle can play the best role to ensure the completion of the task performance.

Most research on occupant workload focus on mental workload. Mental workload is a multidimensional concept, which is closely related to the information processing. Mental workload involves the work requirement, time pressure, task complexity, operator’s ability, operator’s effort and so on [2]. The main measurement methods are director measurement method, sub-task measurement method, subjective evaluation method and physiological measurement method [3].

According to the definition of cognitive load, task load is closely related to information received when the task is executed. Therefore, this paper proposes an objective evaluation method of task load for operational task flow based on multiple resources theory [4], which combines with task performance measurement methods, subjective evaluation methods and physiological measurement techniques. The method studies the change of task load before and after the change of occupant’s function based on vehicle semi-physical simulator, which provides reference for the function allocation of occupant.

2 Method

2.1 Task Design

Long range target attack and close target attack are selected to verify availability of method.

Except for distance to attack target, two types of tasks have difference in operator performed attack operation in the 3-occupant mode. In the long range target attack mode, Gunner is responsible for receiving instructions from commander, aiming at long distance target and attacking target. In the close target mode, commander is responsible for attacking close target [5].

Except for distance to attack target, two types of tasks have difference in attack operation, the attack operation performed by gunner is originally assigned to commander in the 2-occupant mode. The task load of commander is considered much larger than it in 3-occupant mode before experiment. Semi-physical simulator experiment need to be conducted to evaluate task load and determine whether high task load influencing task operation. Commander is responsible for assigning instructions, aiming at long range target, attacking target and reporting to superior in the long range target attack mode. Commander is responsible for attacking close target in the close target attack mode.

2.2 Experiment Design

2-occupant experiment adopts experiment design with two independent variables.

  • Occupant group. between-subjects designed, there are two levels including driver and commander.

  • Task type. within-subjects designed, there are two levels. Every subject in groups performs long range attack task and close attack task, each level repeat twice in turn. The order of task is long range attack task, close attack task, close attack task and long range attack task. Experiment design shows in Table 1.

    Table 1. Experiment design
    Full size table

2.3 Subjects

Subjects are healthy PRV drivers. They have no eye disease and their vision or corrected visual acuity in 1.2 or more. Subjects are in good health condition during the experiment. There are 30 male subjects who are randomly divided into commander and drivers groups averagely.

2.4 Experiment Apparatus

The experiment apparatus include a vehicle semi-physical simulator, a camera, a physiological feedback instrument and a detection response task (DRT) [5]. Table 2 shows their functions.

Table 2. Experiment apparatus and functions
Full size table

The subjective evaluation is carried out after each experiment. The subjects rate overall load and 5-channel load respectively from 0 to 10 in different task type and different occupant type. The 5-channel indicates visual channel, auditory channel, cognitive channel, behavioral channel and oral channel. Higher score represents the heavier load.

2.5 Experiment Process

In order to master the use of experiment apparatus, subjects are trained to be familiar with the operation of DRT equipment and the vehicle semi-physical simulator before the start of the experiment. Then, subjects are asked to perform a number of tasks related to the experiment, and let subjects wear apparatus of physiological and DRT after finishing above operation. The experiment is carried out according to the above experimental design. Report “Attention, task complete, please assign next task” to the lab after completing each task type, and subjects process to the next task experiment when they hear “task continue”, and it is the time to finish the experiment until subjects hear the voice of “End experiment”. After completing each experiment, subjects are asked to mark for subjective rating scale.

3 Results and Discussion

3.1 Data Description

Extract index for the single task type experiment. The load-research index includes two type of physiological index and subjective evaluating data for 5-channel load, one type index include occupant’s primary-task performance (second, s), sub-task (cognitive attention) performance (response time, ms; miss rate, %), the galvanic skin response, and another type is the heart rate variability. Occupant act coherently in experiment, therefore only test commander’s load. Various task-moments during the target attack are depicted as the Fig. 1.

Fig. 1.
figure 1

Each task phase during target attack

Full size image

The time points which are required to record are commander spotting target, commander accepting attack-instruction from superior attacking target successful (when hearing cannon’s sound) and commander restoring driving instruction (including reporting to superior and ordering occupant). Gather commander’s accurately data of attacking target from using camera after the experiment. When it comes to the lack of data results from experiment apparatus and metering mistake, we adopt mean-value invariant principle to substitute the same tested metering data in corresponding metering group. The extra data need to gather show in the below Table 3.

Table 3. Commander’s task load evaluation data
Full size table

3.2 Data Analysis Results and Discussion

The data analysis focuses on the influence for the commander’s load after merging the gunner and commander’s duty. The analysis results are summarized from three aspects. The first aspect focus on the comparative analysis between the task performance of the 3 occupants or 2 occupants and DRT measurement data, and this part mainly discuss the influence of different task types and number of occupant on the task load of commander. In the second aspect, we accomplish a comprehensive analysis about the change of task load in the process of commander from physiological index and subjective score. In the third aspect, the subjective score is taken as the object of analysis, and the relationship between the variation regulation of different channels’ rate with the process of the task and the total load rating is discussed.

Task Performance and DRT Measurement Data Analysis.

The completion time of each task stage in different task type and the quantifiable load value from DRT are shown in Table 4.

Table 4. Task performance and DRT load quantitative statistical results (mean)
Full size table

1. Primary-Task Performance Data Analysis.

From the performance result, the change in the number of occupants has an impact on the different tasks performance. In the second phase that implement the long range target attack, we can see that the commander task performance of the 2-occupant is significantly higher than the task performance of gunner in the 3-occupant (32.5 s < 50.44 s, F(1, 54) = 5.5, p = 0.023 < 0.05). It illustrates that the 2-occupant mode completed faster than the 3-occupant mode in the task, and the performance of commander has been improved in the task phase 1 and phase 3.

The long range target attack is more difficult and time-consuming than the close target attack at the 2-occupant mode. This is not only reflected in the completion time of the attack task, but also in the response time of the commander order task. The time spent by the commander assign task in the long range target attack mode is significantly higher than the close mode (F(1, 53) = 9.36, p = 0.003 < 0.05).

2. DRT Measurement Data Analysis.

From the result of DRT measurement data (Fig. 2), the number of occupants on the influence of commander’s load is associated with task types. The task load of commander is significantly increased in two different task types, and this phenomenon is more obvious in the long range target attack task type. Attacking task is performed by gunner in the 3-occupant mode, and the executor is commander if in the 2-occupant mode. This resulted in the markedly increasing of the task load, which it doesn’t cause any decreasing of the task performance, it improves instead.

Fig. 2.
figure 2

Commander’s DRT load analysis under different occupants (95% confidence interval)

Full size image

The time of target search is under the influence of task type and the number of occupants, as Fig. 3. In close target attack task type, the decreasing of the number of commander result the observably increasing of the search time. In long range target attack task type, the decreasing of the occupant number result the decreasing of the search time. So the interaction between task type and the number of occupant on the influence of target search time is observably. Task type has dominant impact on the target search time, and the search time of long range target is higher than the search time of close target (F(1, 176) = 9.80, p = 0.002 < 0.01) in both two different mode of the occupant’s number.

Fig. 3.
figure 3

The time of target search (95% confidence interval)

Full size image

The Change Law of Physiological Index Data During the Process of Tasks.

The physiological index extract the original heart rate and the GSR (galvanic skin response) signal. The SCL index in GSR signal, the LF/HF index in the HRV (heart rate variability) and the Pulse can effectively represent the degree of the task-performing person’s load. The more value-increasing compared to the baseline suggests the larger load [6].

In order to eliminate the time effect and the individual difference, we use the normalization data to process when analyze the load of task-performing persons in the different task type, and normalization proceed in the commander or gunner’s task type measured data, proceed Z-score standardization on one index of single event time order.

The normalization of commander physical indexes is shown in Fig. 4. The values are standardized indicators, if one is greater than zero, it suggests that it is greater than the average indicator of the whole task phase, if one is less than zero, it suggests that it is less than the average indicator of the whole task phase. The load variation of the whole task represents a high-to-low trend on the time order.

Fig. 4.
figure 4

The normalization of commander physical indexes

Full size image

The Change Law of the Subjective Score of Multichannel Load in Task Process.

The subjective rate obtains not only the total load of every phase in the tasks, but the visual, auditory, cognitive, behavioral and oral load. Under the different tasks, the summarize of relevancy between the score of total load and each channel is shown in Table 5. The relevancy identified by Pearson Correlation. As a result, there has a reasonably high correlation between the score of task load and the score of visual, cognitive and behavioral channels, particularly in the long range target attack task type. Above mentioned correlation increase when the executor from gunner shifted to commander, which the score of behavioral channel under the significant influence of the number of occupant (F(1, 30) = 4.16, p = 0.051). In close target attack task type, 2-occupant mode has a more obvious decreasing of load score in the cognitive channel than 3-occupant mode.

Table 5. Correlation between task load score of commander and total load score in different channels
Full size table

4 Conclusion

This paper adopts a experimental method which based on human-in-loop simulator to research the task load change because of the change occupants’ function. Occupants’ task load (in two task types) is evaluated by task performance measurement, subjective estimate method and physiological quantitative measurement technique. We decompose the task types and design subjective evaluation questionnaires according to cognitive channel. We also analyze the result of task performance, subjective evaluation and physiological measurement, and obtain the following conclusions:

The task load of commander in 2-occupant mode is significantly higher than the 3-occupant, but it doesn’t decrease the performance of task. On the contrary, the completion time in 2-occupant is less than 3-occupant. The performance of commander is also promoted while in task phase 1 and phase 3. For the long range target attack task type which is more difficult, the time for commander searching target is significantly decreased.

The impact of occupant number change on of subjective evaluation of commander’s overall task load is mainly reflected on the task type of long range target strike task in 3-occupant mode, the subjective load score is high in the middle and low in both ends and long distance target is difficult to strike, as a result, the score of the attacking task performance is significantly higher than the other phase. In 2-occupant and 3-occupant mode, the type for close range target strike, are performed by commander attack task. Therefore, in different number of occupant modes, the change of subjective score tends to be consistent.

The total score of task load is highly related to vision, cognition and behavior channel rate. The correlation is increasing when the executor from gunner shifted to commander, particularly in the long range target attack task type. The commander’s channel load score in 2-occupant mode is significantly higher than the 3-occupant mode. Compared the type of close range target task between 2-occupant mode and 3-occupant mode, the score of cognitive channel load decreased significantly in 2-occupant mode.

The conclusion of this study can be made that the change of the occupant number has a greater impact on the occupant performance with a increased physical load within the acceptable range; The evaluation method based on task load is feasible in the semi-physical simulator environment. All in all, the study can provide theoretical support for the evaluation of the occupant assignment function in the new type of PRV. The evaluation method of this study can also be used to other manned operating systems, such as aircraft, ship, and spacecraft and so on.