TY  - THES
AB  - The field of human-robot interaction deals with robotic systems that involve
humans and robots closely interacting with each other. With these systems
getting more complex, users can be easily overburdened by the operation
and can fail to infer the internal state of the system or its ”intentions”. A
social robot, replicating the human eye region with its familiar features and
movement patterns, that are the result of years of evolution, can counter
this. However, the replication of these patterns requires hard- and software
that is able to compete with the human characteristics and performance.
Comparing previous systems found in literature with the human capabili-
ties reveal a mismatch in this regard. Even though individual systems solve
single aspects, the successful combination into a complete system remains
an open challenge. In contrast to previous work, this thesis targets to close
this gap by viewing the system as a whole — optimizing the hard- and
software, while focusing on the replication of the human model right from
the beginning. This work ultimately provides a set of interlocking building
blocks that, taken together, form a complete end-to-end solution for the de-
sign, control, and evaluation of a human-inspired robotic eye. Based on the
study of the human eye, the key driving factors are identified as the success-
ful combination of aesthetic appeal, sensory capabilities, performance, and
functionality. Two hardware prototypes, each based on a different actua-
tion scheme, have been developed in this context. Furthermore, both hard-
ware prototypes are evaluated against each other, a previous prototype, and
the human by comparing objective numbers obtained by real-world mea-
surements of the real hardware. In addition, a human-inspired and model-
driven control framework is developed out, again, following the predefined
criteria and requirements. The quality and human-likeness of the motion,
generated by this model, is evaluated by means of a user study. This frame-
work not only allows the replication of human-like motion on the specific
eye prototype presented in this thesis, but also promotes the porting and
adaption to less equipped humanoid robotic heads. Unlike previous systems
found in literature, the presented approach provides a scaling and limiting
function that allows intuitive adjustments of the control model, which can
be used to reduce the requirements set on the target platform. Even though
a reduction of the overall velocities and accelerations will result in a slower
motion execution, the human characteristics and the overall composition of
the interlocked motion patterns remain unchanged.
DA  - 2020
DO  - 10.4119/unibi/2943734
LA  - eng
PY  - 2020
TI  - Design, Control, and Evaluation of a Human-Inspired Robotic Eye
UR  - https://nbn-resolving.org/urn:nbn:de:0070-pub-29437348
Y2  - 2024-11-22T05:01:48
ER  -