Abstract:
Unmanned aerial vehicles (UAVs) have emerged to be a powerful tool in several industries.
Moreover, there is increased interest in aerial manipulation tasks such as canopy sampling.
Aerial manipulation uses UAVs to perform interaction tasks with the environment. Several
applications are safer and more efficient when replacing human workers with UAVs. Canopy
sampling, for example, requires obtaining tree samples from forests, which would take far
longer and be significantly more challenging for human workers. Other aerial manipulation
application examples include cleaning up dangerous sites, such as after a natural disaster
or for nuclear facilities. An essential aspect of aerial manipulation is dexterity, which
is generally used to describe the ability of human or robotic hands to perform tasks.
Currently, there is limited research on the dexterity of UAVs. Furthermore, no existing
standard tests can be used to measure the dexterity or aerial manipulation capability of
fully actuated UAVs without complex manipulator arms. This research looks to develop
a set of standardised tests to evaluate a fully actuated UAV’s ability to perform aerial
manipulation.
Various aspects of this research have been bio-inspired. Like the robotic arm and hand
dexterity research, which is inspired by human hands, this research has been influenced
by one of nature’s aerial manipulators, the hummingbird. This bird species can hover and
perform complex aerial manipulation tasks. The typical motions and tasks of hummingbirds
have been integrated into this research.
This work explores the concept of dexterity in the context of UAVs and investigates how
UAV dexterity can be quantified. A definition of UAV dexterity has been synthesised with
existing knowledge of dexterity in human and robot arm or hand dexterity. Additionally,
by exploring the range of existing standardised testing methodologies, testing requirements
are established to describe the elements of a suitable standardised dexterity test. During
this research, three standardised UAV dexterity tests are developed and experimented with.
A reachability test has been designed and tested experimentally. The proposed test evaluates
a UAV’s ability to reach a point in space while station keeping. The presented work
describes a formula that leads to a reachability score. Several experiments are conducted
to produce data and a reachability score for a fully actuated octocopter. A dexterity test
inspired by the dexterity concept of force closure is also presented. This explores interaction
and how a UAV capable of applying a sustained force is more dexterous. Finally, a
peg-in-hole test is developed, which measures a UAV’s ability to perform precise aerial
manipulation tasks without the complications of interaction. This test is heavily inspired by
human peg-in-hole tests and hummingbird’s everyday activities involving collecting nectar
from slender flowers, imitating a peg-in-hole action. These tests have been experimented
with to show the feasibility of assessing UAV dexterity