We analyze the work time and mental workload for each group in each day to measure the total effect of AR prompts support. Both of them are evaluated by average. We also compared the moving time, grasping time and transporting time, releasing time and fre-quency of collision to evaluate the effect of each AR item.
5.4.1 Accomplishment time
We compared the average task time in all environments for each group and derived the following graph (Fig. 6 (a)). In the first day, all conditions and tasks of two groups are the same. The average task time of group B is worse than that of group A, which should be nearer as expected. How-ever, in the second day, the average task time of group B is better than that of group A. Additionally, the difference of average task time of group B in two days is extreme significant. So, we can easily know that the AR prompt can raise the time efficiency.
Figure 5.6 Work time in different situations 5.4.1.1 In base movement
Figure 5.6 shows the average time used in base movement condition in all tasks of each group in each day. In the first day, time used of group B is a little more than that of group A. However, in the second day, time used of group B decreased obviously while that of group A decreased only a little. The difference of moving time of group B in 2 days is extreme significant, so that we think the reachable sphere can help operator to find a position where they can conduct the following tasks well.
5.4.1.2 In grasping
In this period, operators conduct the most subtle manipulation to grasp debris cylinder.
Any careless manipulation may cause grasping failure and increase the time cost. Ac-cording to Figure 5.6, average time cost of two groups in the first day are almost the same. However, in the second day, the average grasping time of group B decreases more obviously than that of group A. The difference of average grasping time of group B in two days are extreme significant. So, we can easily know that the guide laser can reduce the grasping time by informing operators the relationship between two pieces of end-effector and other objects.
5.4.1.3 In transport
After grasping debris, operator should transport it to recycle box. During this period, operators tend to rotate in the opposite direction as expected at the beginning. They also
0 100 200 300 400 500 600
***: 𝑝<0.001
0 50 100 150 200 250
Time s
Based
movement Grasping Transport Releasing Overall
Work time
***
***
***
***
***
Group A Group B
Day 1 Day 2 Day 1 Group B Day 2
Without support With support
tend to rotate the machine upper part in a worse direction, which may cause the total rotation time much more than expected. Through the rotation hints, it seems operators are easily understand the relationship be-tween control lever and detail view of manipu-lator, which is indicated in Figure 5.6. Group B performed worse than group A under the same condition in the first day. However, they did almost the same as group A did in the second day. The difference of the transporting time of group B in two days are ex-treme significant, so that we can derive that the rotation hint can reduce the time cost in transporting period.
5.4.1.4 In releasing
According to Figure 5.6, group B cost much more time when releasing debris than group A in the first day. In the second day, group B performs improves much more than group A. The time cost of two groups in releasing period are almost the same in the second day. Time difference of group B in two days is extreme significant. It means that vertical arrow can increase the time efficiency in re-leasing by judging whether the ar-row is in the recycle box instead of imagining relationship between debris and recycle box.
5.4.2 Mental workload
Here, we used NASA-TLX method in our questionnaire to measure the mental work-load [43] [73] [74]. Mental demand (MD), physical demand (PD), temporal demand (TD), own performance (OP), frustration (FR) and effort (EF) are scored from 0 to 100 by operators. The higher the heavier. They are also asked to compare the importance between each 2 items from the above 6 to measure the weight of each item, which are scored from 0 to 5 and the total weight is 15. So the WWL score is given by [∑(𝑆 ⋅ 𝑊)] 15⁄ . Here, 𝑆 stands for score of each item and 𝑊 stands for weight of each item. Such WWL score is used to measure each operator’s mental work directly. The higher the score is, the heavier the mental workload is.
According to the calculation, the average mental workload of each group in each day is shown in Figure 5.7. It is easy to find, there is almost no difference between two groups in the first day. However, in the second day, the average mental work of group B de-crease obviously while that of group A keeps the same level as the first day. So, we think the AR vision prompts can reduce the operators’ mental workload. It also means that the AR prompts can supply a better user experience.
Figure 5.7 Average mental workload of each group
5.4.3 Frequency of collision
Figure 5.8 show the average number of collision times in all tasks of each group in each day. In the first day, the collision frequency of group B is twice than that of group A.
However, in the second day, collision frequency of group B decreases obviously while that of group A keeps at a stable level as the first day. That means the distance arrow in our AR vision prompts can reduce the possibility of collision to some extent.
Figure 5.8 Average error contact per task of each group
Group A Group B
Day 1 Day 2 Day 1 Group B Day 2
Without support With support
0 10 20 30 40 50 60 70 80 90
Mental workload
WWL score
Group A Group B
Day 1 Day 2 Day 1 Group B Day 2
Without support With support
Number of error contact
0 0.5
1 1.5
2 2.5
3
Number of times
*: 𝑝<0.1
*