IEEE Systems, Man and Cybernetics Magazine - October 2020 - 18

(m)

1.5

0.5

-0.5

sition

-1

Y Po

-1.5

0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
1.5
1
0.5
0
-0
.5
-1
-1
.5
-
-2 2
.5

X Position (m)

Real Tracker Position
Vive Estimated Position (Open Source)
Lighthouse Position
Figure 3. A 3D plot of the actual positions and the

positions estimated by the Vive tracker using opensource software.

Localization Using Proprietary HTC Vive Software
To test the performance of the localization using the
closed-source HTC Vive software, the same experiment as
in the "Localization Using Linux-Based Open-Source Software" section was conducted [30]. As shown in Figures 4
and 5, the closed-source software performs better than the
open-source software. The problem with radial drift still
occurs, but it is significantly less than the open source's
radial drift error.
Long-Term Localization
For the next experiment, a UGV was manually driven in
a square with VIVEPOSE tracking the absolute position
of the UGV. This was repeated five times and at each of
the corners of the square the actual, odometry-estimated, and Vive tracker-estimated positions were recorded.
Figure 6 shows the Euclidean distance between the actual and odometry-estimated positions plotted as a solid
green line and the Euclidean distance between the actual and VIVEPOSE estimated positions plotted as a solid
red line. The average odometry-estimated positional
error trend line (represented by the dashed green line) is
increasing significantly over time, but the VIVEPOSE's
positional error trend line (represented by the red
dashed line) remains almost constant. This means that
the VIVEPOSE positional error is not cumulative and,
therefore, does not increase over time.
Predefined Path Following
In another experiment, a proportional-integral-derivative
(PID) controller was used to drive the UGV in two different
test patterns, a square and a figure eight. The PID controller used the odometry sensor's positional information as

1.5

0.65
0.6
0.55

Y Position (m)

1
0.5

0.5
0.45

0
-0.5

0.4
0.35
0.3

-1
-1.5

Real Tracker Position
Vive Estimated Position (Proprietary Software)
Lighthouse Position

Figure 4. A 2D plot of the actual positions and the

positions estimated by the Vive tracker using the
proprietary HTC Vive software.

IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE O ctober 2020

X Position (m)

-1
-0.5
0
0.5
1
1.5

1.5

-2
-1.5

-2.5

-0.5
0
0.5
X Position (m)

-1

0.25

0
0.
5

-2.5
-1.5

-1
.5
-1
-0
.5

-2

Z Position (m)

respect to the true ground plane. This problem remains
with two lighthouses and the original software interface. Therefore, the most likely cause of the radial drift
is the anisotropic properties of the emitted light pattern at the bounds of the FOV, which was also characterized by Niehorster et al. [29] as "deviations from
isotropy were mostly small except at some positions
along the edge of the grid." In the case of our formation-control experiment, the anisotropic effect will be
more pronounced as the lighthouse is closer to the
ground plane.

Y Position (m)

Real Tracker Position
Vive Estimated Position (Proprietary Software)
Lighthouse Position

Figure 5. A 3D plot of the actual positions and the

positions estimated by the Vive tracker using the
proprietary HTC Vive software.

IEEE Systems, Man and Cybernetics Magazine - October 2020

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - October 2020