IEEE Robotics & Automation Magazine - December 2017 - 127

HUMANITARIAN TECHNOLOGY

Thethe
On
2017
Ethics
Humanitarian
of Research
Robotics
and Practice
and in
Robotics andTechnology
Automation
Automation
Challenge
By Raj
Alexandre Amory, Edson Prestes, Renan Guedes, Augusto Bergamin, Renata Neuland,
RajaMadhavan,
Chatila
Mathias Mantelli, Diego Kindin, and Fernanda Rodrigues

ince 2014, the IEEE Robotics
and Automation Society's
Special Interest Group on Humanitarian Technology (RASSIGHT) has been organizing the
Humanitarian Robotics and Automation Technology Challenge (HRATC).
The main goal of the HRATC is to
develop reliable robotic solutions for
detecting landmines and unexploded
ordnance, especially for communities
where, due to these explosives, people
live in fear of losing limbs or even their
lives. Consequently, these communities
have reduced access to agricultural
lands for growing crops, which severely
restricts their sustenance and livelihood.
While other solutions exist, they are
prohibitively expensive and are not
practical for the developing world,
where the bulk of these mines are buried. The emphasis of the HRATC from
the beginning has been the development of cost-effective and sustainable
solutions for such communities. More
details on previous editions of the challenge are available in [1]-[3].
In the 2017 edition of the HRATC,
one of the objectives was to design an
affordable robot that could be built and
deployed for mine detection. To that
end, the first change in the robot was
the utilization of a Pioneer P3-AT
mobile base instead of the Clearpath
Husky used in past editions. The robot,
shown in Figure 1, consists of the Pioneer P3-AT base and fixtures used to
keep the electronics fully described on
Digital Object Identifier 10.1109/MRA.2017.2757722
Date of publication: 13 December 2017

Figure 1. The HRATC 2017 robot platform
and sensor suite.

the RAS-SIGHT GitHub [4]. The second physical modification to the robot
compared to the previous edition is the
selection of lower-cost sensors for localization and obstacle detection, as these
are two of the main technical challenges. The robot's sensor suite consisted of two laser rangefinders: a SICK
Tim-551 facing forward, for localization, and a Hokuyo URG-04LX positioned below the SICK rangefinder, for
detecting obstacles; an MTK3339-based
GPS module and a CH Robotics UM6
inertial measurement unit were for
positioning. A custom robot operating
system (ROS) package was built with all
the necessary files for start-up and operation of the robot (the hratc_robot package [4]) and was executed remotely in a
Raspberry Pi 2 Model B through a laptop computer connected to the robot's

Wi-Fi network. The third physical
change compared to the other HRATC
editions is that the metal detector
(attached to the front of the mobile
base, as shown in Figure 1) was built
from scratch with commonly available
parts based on an existing design [5].
Similar to past editions of HRATC,
the teams' main challenge was to successfully translate their solution from
the simulation phase to a real-world
environment during the testing phase.
The participating teams were required
to develop their solutions in the ROS
framework, which would then run in
the laptop after starting up the robot.
The tests were performed in a controlled area of approximately 360 m2,
where mock mines were buried. The
robot localization challenge was simplified in the past iterations of the
HRATC because the robot used a realtime kinematic GPS and the test scenario was on even terrain in an open
field with good GPS availability. However, in this edition, the robot used a
simpler GPS module, and the scenario had occlusions from trees and
a nearby building. Though a more
realistic scenario, this proved to be
challenging for the teams, who had to
develop solutions not completely reliant on GPS. Moreover, the terrain
irregularities caused the laser rangefinders to sway, compounding errors
in localization.
The participating teams integrated and tested several localization
techniques based on sensor fusion, e.g.,
extended Kalman and particle filters
and simultaneous localization and

DECEMBER 2017

IEEE ROBOTICS & AUTOMATION MAGAZINE

127

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2017