IEEE Robotics & Automation Magazine - March 2022 - 18

Table 1. The alignment accuracy results, showing the positional difference among the holographic
and actual three desired robot end-effector locations. All the dimensions are shown in millimeters.
Point 1
Sample TX
1
2
3
4
5
TY
−6.4
2.64
TZ
Norm TX
−7.43 −13.35 13.99 20.72 0.14
−0.85 1.52
5.34
4.73
3.16
−12.15 4
−1.86 2
TY
−3.44 −7.83 10.67 −2.24 5.95
4.77
Point 2
TZ
−11.57 5.79
9.71
13.87 −5.59 8.94
5.46
Norm TX
TY
Point 3
TZ
Norm
Average
Euclidean
Norm
−9.84 11.71 −1.58 −1.08 -6.35 12.87 11.76
12.94 −6.18 −4.1
−1.27 10.89 −4.17 5.61
6.91
−0.41 −1.64 9.06
based on a desired end-effector position captured through
the virtual environment were calculated using the method
presented in [7], which consists of sampling reconfigurations
that satisfy the end-effector vector, defined by P5
and 6 generated through distance geometry. From these
optimal reconfigurations, the new position of the distal
link was generated and displayed in the virtual environment
as a holographic distal link. Each configuration was
reported in terms of actual interpoint distances (D13,
D14, D23, and D24, as depicted in Figure 2) and compared
with the desired interpoint distances to compute
the positioning difference ().T By denoting the actual distances
with subscript a and the desired distances with
subscript d, the reconfiguration error is described as
T =-13 13 13
DD D
DD
d=+ ++
`
4
1
D13
13
D14
D23
D23
and so on for ,,14 23TT and TD .24
The percentage errors were also calculated to provide context
for the absolute errors relative to the interpoint distances
D
T TTTD24
D14
D24
100
90
80
70
60
50
40
30
20
10
j
(8)
da
(7)
3.3
7.73
12.61 −7.79 2.41
9.21
6.54
−0.76 −0.75 0.06
12.75 14.75 16.13
7.26
10.45 12.31
1.07
5.25
It may be noted that the length of the distal link could
affect alignment accuracy, as during a reconfiguration it is
aligned to a holographic model of itself, and therefore a
longer distal link would produce larger visual misalignments,
making it easier for a user to achieve a higher alignment
accuracy. We therefore evaluated the effect of the
length of the distal link on alignment accuracy. An identical
reconfiguration was performed using different lengths
of distal link (0-300 mm in steps of 100 mm), each repeated
five times to obtain an average. For each of these reconfigurations,
the error in interpoint distances (D13, D14,
D23, and D24) was computed, and the results are displayed
in Figure 9.
Experimental Setup
Five male participants between the ages of 20 and 26 and
experienced in robotics but with no experience regarding
manipulating malleable robots, were then invited to align
the distal link to the desired location viewed in the HoloLens.
This was carried out as a service evaluation and
therefore did not require ethics approval. Each participant
was required to familiarize himself with the HoloLens and
its control using the built-in HoloLens introduction app
prior to the experiment, where all participants were asked
to complete the alignment process within the allocated
5 min. During this stage, each user is prompted to carry out
the built-in calibration process, where the HoloLens utilizes
the users' interpupillary distance to correct for hologram
offsets resulting from wearing the HoloLens differently. At
the end of the experiment, once the users are satisfied with
their alignment, or if the 5-min time frame has elapsed, the
real robot's configuration is locked and subsequently measured
using the aforementioned OptiTrack motion-tracking
cameras. These results for each sample and user are
shown in Table 2.
Following the reconfiguration experiment, each user was
Component
Figure 9. Results on the effect of different distal-link lengths
on the reconfiguration accuracy, represented by the error in
interpoint distances D13, D14, D23, and D24 of the alignment.
18 * IEEE ROBOTICS & AUTOMATION MAGAZINE * MARCH 2022
asked to complete the NASA-Task Load Index (TLX) questionnaire.
Developed by the NASA Ames Research Laboratory
for subjective evaluation of task workload, the
NASA-TLX is a widely used, subjective, and multidimensional
survey that measures the relative contributions to
NASA-TLX Scale
Mental
Workload
Physical
Workload
Temporal
Workload
Performance
Effort
Frustration
IEEE Robotics & Automation Magazine - March 2022

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2022
