IEEE Robotics & Automation Magazine - December 2022 - 148

COMPETITIONS
Autonomous Ground Navigation
in Highly Constrained Spaces
Lessons Learned From the Benchmark Autonomous Robot Navigation Challenge at ICRA 2022
By Xuesu Xiao , Zifan Xu, Zizhao Wang, Yunlong Song, Garrett Warnell, Peter Stone,
Tingnan Zhang, Shravan Ravi, Gary Wang, Haresh Karnan, Joydeep Biswas, Nicholas Mohammad,
Lauren Bramblett, Rahul Peddi, Nicola Bezzo, Zhanteng Xie, and Philip Dames
T
he Benchmark Autonomous
Robot Navigation (BARN)
Challenge took place at the 2022
IEEE International Conference
on Robotics and Automation (ICRA),
in Philadelphia, PA, USA. The aim of
the challenge was to evaluate state-ofthe-art
autonomous ground navigation
systems for moving robots through
highly constrained environments in a
safe and efficient manner. Specifically,
the task was to navigate a standardized
differential drive ground robot from a
predefined start location to a goal
location as quickly as possible without
colliding with any obstacles, both in
simulation and in the real world.
Five teams from all over the world
participated in the qualifying simu -
lation competition, three of which were
invited to compete with one another
at a set of physical obstacle courses
at the conference center in Philadelphia.
The competition results suggest that
autonomous ground navigation in
highly con strained spaces, despite
seeming simple for experienced ro -
boticists, is actually far from being
a solved problem. In this article,
we discuss the challenge,
the ap -
proaches used by the top three winning
teams, and lessons learned to direct
future research.
BARN Challenge Overview
Designing autonomous robot navigation
systems has been a topic of interest
to the robotics community for decades.
Digital Object Identifier 10.1109/MRA.2022.3213466
Date of current version: 2 December 2022
Indeed, there currently exist many such
systems that allow robots to move from
one point to another in a collision-free
manner [e.g., open source implementations
in Robot Operating System (ROS)
that have extensions to different vehicle
types], which may create the perception
that autonomous ground navigation is a
solved problem. This perception may be
reinforced by the fact that many mobile
robot researchers have moved on to
orthogonal navigation problems
beyond the traditional metric (geometric)
formulation and that focus only on
path optimality and obstacle avoidance.
These orthogonal problems include,
among others, learning navigation systems
in a data-driven manner, navigating
in off-road and social contexts, and
multirobot navigation.
However, autonomous mobile robots
still struggle in many ostensibly simple
scenarios, especially during real-world
deployment. For example, even when
the problem is simply formulated as traditional
metric navigation so that the
only requirement is to avoid obstacles
on the way to the goal, robots still often
get stuck and collide with obstacles
when trying to navigate in naturally
cluttered daily households; in constrained
outdoor structures, including
narrow walkways and ramps; and in
congested social spaces, such as classrooms,
offices, and cafeterias. In such
scenarios, extensive engineering effort is
typically required to deploy existing
approaches, and this requirement presents
a challenge for large-scale unsupervised
real-world robot deployment.
Overcoming this challenge requires sys148
* IEEE ROBOTICS & AUTOMATION MAGAZINE * DECEMBER 2022
tems that can both successfully and
efficiently navigate a wide variety of en -
vironments with confidence.
The BARN Challenge was a competition
at ICRA 2022, in Philadelphia,
that aimed to evaluate the capability of
state-of-the-art navigation systems to
solve the previously mentioned challenge,
especially in highly constrained
environments where robots need to
squeeze between obstacles to navigate
to the goal. To compete in the BARN
Challenge, each participating team
needed to develop an entire software
stack for navigation for a standardized
and provided mobile robot. In particular,
the competition provided a Clearpath
Jackal with a 2D 270º field-of-view
Hokuyo lidar for perception and a differential
drive system with 2-m/s maximum
speed for actuation. The aim of
each team was to develop the navigation
software stack needed to autonomously
drive the robot from a given starting
location through a dense obstacle field
to a given goal and to accomplish this
task without any collisions with obstacles
and any human interventions.
The team whose system could best
accomplish this task within the least
amount of time would win the competition.
The BARN Challenge had two
phases: a qualifying phase evaluated in
simulation and a final phase evaluated
in a set of physical obstacle courses. The
qualifying phase took place before the
ICRA 2022 conference, using the
BARN data set, which is composed of
300 obstacle courses in Gazebo simulation
and randomly generated by cellular
automata. The top three teams from the
http://orcid.org/0000-0001-5151-2186
IEEE Robotics & Automation Magazine - December 2022

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