IEEE Systems, Man and Cybernetics Magazine - October 2020 - 17

Further filtering and optimization for faster and more
reliable localization can be performed using linear fit
methods like random sample consensus or sensor fusion
methods such as EKF of the position estimates from the
onboard inertial measurement unit on the tracker.

However, as Niehorster et al. [29] determined from
experiments using a SteamVR plugin, the head-mounted device (equipped with the same sensors) is still
affected by a reference plane that is slanted with

Results
To test the accuracy of the VIVEPOSE system, several
experiments were performed, including localization using
Linux-based open-source software, localization using proprietary HTC Vive software, long-term localization, and
predefined path following.
Left Sweep Extreme

HTC VIVE
Lighthouse
(IR Laser
Projector)

Optical Axis

Sensor

Right Sweep Extreme

Figure 1. The top view of the HTC VIVE Lighthouse,

showing interactions with a sensor.
1.5
1
0.5
Y Position (m)

Localization Using Linux-Based Open-Source Software
The first experiment tested the localization of the VIVEPOSE system using open-source software in Linux [28].
The benefits of using open-source software is the ability
to combine the tracking software with other software,
such as the robot operating system (ROS). However, the
open-source software may not be as well developed as the
proprietary software that comes with the device, so it
could result in lower accuracy. To test the accuracy of the
open-source software, we placed the Vive tracker at various positions and recorded the actual and VIVEPOSE estimated positions. In the test, the lighthouse is about 0.6 m
and at a 45° incident angle from the ground floor. The
results of the experiment can be seen in Figures 2 and 3.
A noticeable radial drift occurs as the tracker's location
is away from the calibrated (0, 0, 0) position, likely due to
the nature of the radial sweeping pattern and lighthouse
pose in our tests. Calibrating the device with a single lighthouse was initially thought to be the cause of the radial
drift. Calibration routines for the open-source software
requires one or two lighthouses and that the Vive tracker
initially remain still on the ground for some time. This contrasts greatly with the SteamVR calibration process, which
requires the following:
◆ Two lighthouses are mounted high above and far away
from the user.
◆ Lighthouses are oriented so that there is a line of sight
between them.
◆ The user establishes a ground level by placing a Vive
tracker on the floor.
◆ The user defines bounds of the area for which the system is used (basically "draws a box" by walking around
the FOV with the handheld controller).
◆ No objects are between the user's controllers and the
lighthouses.
With the SteamVR constraints on the placement of the
lighthouse IR beacons, the effects of a radial drift pattern
are likely reduced by
◆ providing optimal coverage patterns
◆ yielding more measurements for the tracker per
lighthouse
◆ providing a secondary set of data points from a different angle to correct the radial drift pattern.

0
-0.5
-1
-1.5
-2
-2.5
-1.5

-1

-0.5
0
0.5
X Position (m)

1.5

Real Tracker Position
Vive Estimated Position (Open Source)
Lighthouse Position
Figure 2. A 2D plot of the actual positions and the

positions estimated by the Vive tracker using opensource software.

O c tob e r 2020

IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE

IEEE Systems, Man and Cybernetics Magazine - October 2020

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