IEEE Robotics & Automation Magazine - September 2023 - 156

to change the layout of the arena to better
test the robots' and teams' capabilities
in challenging environments and
maximize visibility for the audience.
Two weeks before the event, the local
organizers finalized the arena layout
and shared the setup,
including changes to the
grading policy, with the
teams. Although this
was a late change for the
teams, the main idea of
testing the quadruped
robots' capabilities in
handling challenging
environments remained
unchanged.
The new arena was
"
divided into five sections:
ramp terrains, soft
floor, pallet steps with
pipes, K-rails, and negative
steps. Each section
had a starting area and
an ending area separated
by a screen from
other sections, allowing
teams to run their robots
in each section without
interference from others during testing
or the game. During the qualifying
rounds, all five sections were populated
with the teams' robots to make full use of
IN SUBSEQUENT
STAGES IN A FEW
YEARS, AS QUADRUPED
ROBOTS
WITH MANIPULATORS
BECOME
MORE PREVALENT
AND ACCESSIBLE
TO MORE TEAMS,
PRACTICAL MOBILE
MANIPULATION
TASKS CAN BE
INTRODUCED.
.
„
Crossing Ramps: Flat
Soft Foam: Flat
the arena and provide more to show the
audience. If teams ran the entire semicircular
track one by one, only a part of
the arena would be utilized, resulting
in wasted time and space. Additionally,
by having multiple rounds of qualifying
events, teams had more
opportunities to adapt
their technologies to the
environment and progressively
improve their
algorithms if necessary.
The arena was constructed
by a team from
the National Institute of
Standards and Technology,
led by Adam Jacoff,
and Oliver Huke from
the U.K. Atomic Energy
Authority, Remote
Applications in Challenging
Environments
(RACE) Test Facility.
The difficulty of the
terrain can be adjusted
by changing the slope
in sections A, B, and
D as well as by adding
more obstacles to create
narrower and more complex passages
(Figures 2 and 3). Section A features an
inclined crossing ramp terrain, providing
a reproducible level of complexity
Flat/Easier Lane Settings
Pallet Steps: Straight
with slippery surfaces akin to dust-covered
concrete. In section B, there is a
foam floor with step-overs, which is
designed to simulate walking on surfaces
like sand and water that absorb
the robot's
feet below the perceived
ground plane. Section C involves elevation
changes using covered pallets with
rolling pipe edges to step over, known to
be challenging for tracked and wheeled
robots. Section D replicates a slippery
floor covered with collapsed objects,
represented by diagonal rails. Finally,
section E includes positive and negative
obstacles (holes) that are difficult for the
robot to perceive.
For the remote control category, the
team's operator must control their robot
remotely, relying solely on the scene transmitted
from the robot's onboard sensors
while being out of sight of the lane. All
situational awareness must come through
the operator interface. During the trial,
no additional information is allowed for
the operator except for resetting the robot,
which concludes the current trial and
places the robot back at the starting area
of the section. Regarding autonomy, once
the operator initiates the robot, no further
interaction is permitted. Teams get a
point for a successful run from the start to
the end in the given section. To encourage
teams to use autonomy, our scoring
K-Rails: Flat
Crates: Diagonal Gap
(a)
(b)
(c)
Sloped/Harder Lane Settings
Crossing Ramps: Sloped Soft Foam: Sloped Pallet Steps: Offset
K-Rails: Sloped
Crates: Diagonal Hill
(d)
(e)
(a)
(b)
FIGURE 3. A reconfigurable setup was used to vary difficulties.
15 IEEE ROBOTICS & AUTOMATION MAGAZINE SEPTEMBER 2023
6
(c)
(d)
(e)
IEEE Robotics & Automation Magazine - September 2023

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2023
