IEEE Robotics & Automation Magazine - December 2013 - 76

Table S5. Transition versatility weights.

Table S3. Aerial versatility weights.
Subdomain

N/A

Weight

Mobility

Weight

Mobility

Weight

0.14

T to S

0.14

T to A

0.14

S to T

0.14

S to A

0.14

Aq to A

0.14

A to T

0.14

A to S

0.14

N/A

0.14

Table S4. Aquatic versatility weights.
Subdomain

Surface

Subaquatic

Weight

0.5

Mobility

Weight

0.17

0.14

Table S1 shows terrestrial locomotion. A slightly higher
weight is given to maneuverability based on the assumption
that it is more essential than negotiating obstacles. The
subdomains for maneuverability are given equal weighting.

obstacles. An example would be PackBot, which can use its
front flippers to pitch its body up.
Versatility Grading Methodology
The complete versatility structure is shown in Figure 3, which
includes maneuverability in all domains, obstacle mobility in
terrestrial and scansorial environments, as shown in Figure 2,
and transitions among domains. The leaves represent either the
maneuverability, obstacle mobility, or transition capabilities of
the robot. Each leaf is weighted according to the perceived
importance (see "Versatility Weights"). Each leaf also receives a
score of 0, 1, or 2 based on the robot's capability. A score of 0
76

IEEE ROBOTICS & AUTOMATION MAGAZINE

DECEMBER 2013

Subdomain

N/A

Weight

N/A

For smooth terrain, the most weight is given for forward
motion (x direction). Smaller weights are given for motion in
the y or z directions. No weight is given for roll ( a ) or pitch
( b ) as these features do not provide any advantage on
smooth terrain. Weight is given for yaw ( c ) as this is critical
to turning. Finally, no weight is given for hovering (h). For
obstacle mobility, each obstacle type (obstacle, continuous
obstacle, gap, and slope) is given equal weighting. All obstacle
height categories are given equal weighting. All gap lengths
are given equal weighting. All continuous obstacle heights
are given equal weighting. Slope traversal for slopes is < 30c
weighted slightly higher than the other two slope categories.
Table S2 shows the scansorial weights. Maneuverability
and obstacle avoidance are given equal weighting. Within
maneuverability, rough surfaces are weighted slightly higher
because these are more common. Curved surfaces are given
the least amount of weight. Moving up a wall and changing
heading are given more weight than moving laterally, pitching,
or rolling. No weight is given in the z direction because most
systems should be able to do this by default. Finally, no weight
is given for hovering or perching on the wall.
Aerial weights are shown in Table S3. All maneuverability
components are given equal weighting. In the aerial domain,
the ability to hover is no longer trivial and thus is given the
same weighting as the other maneuverabilities. In addition,
there are no obstacles in the aerial domain.
Versatility weights for the aquatic domain are shown in
Table S4. Equal weight is given for both subaquatic and surface
mobility. Hovering (maintaining a constant position) on the
surface of the water is given no weight. All other motions are
given equal weight for a particular subdomain.
Table S5 shows the transition weights, which are all equal.
Transitions into the aquatic domain are not considered as we
assume all robotic systems have this ability.

means that the robot was not able to accomplish the task. A
score of 1 infers that the robot has a rudimentary capability to
execute the given maneuver, traverse the given obstacle, or
transition among the two given domains. A score of 2 indicates
high proficiency in that particular skill. Scoring is done by
examining the documented capabilities of the robot and
inspecting videos of the robot's performance, where available.
In addition to the weights of each leaf, weights are given to
the middle nodes (subdomains) in the versatility tree, which
represent specific terrains. Again, the weights are chosen based
on the perceived importance to versatility. Thus, the total versatility of a mobile robot is given as

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - December 2013