Analysis and
implementation of control alternatives for underactuated artificial
hands based on the human grasp.
Abstract:
The development of
anthropomorphic and functional artificial hands remains a complex
challenge for design and control engineering. To date, a wide
variety of artificial hands, both robotic and prosthetic, have been
developed. Some designs are aimed at achieving a grip stability and
manipulation skills similar to that of the human hand while others
have been designed to achieve a better anthropomorphism and cosmetic
appearance. However, an artificial hand should combine both
objectives, particularly if oriented to hand replacement.
Unfortunately, current hand prostheses are very simple from a
biomechanical point of view. The fundamental problem lies in the
difficulty of establishing a proper communication between the user
and the prosthesis. There is no simple way to achieve the
interaction between the amputee and a dexterous prosthesis of
multiple degrees of freedom, such as those developed in the recent
decades, since it requires either too many independent control
signals or an intelligent controller. The trends in the control of
robotic and prosthetic hands indicate that a novelty in this field
could be achieved by observing and imitating the natural
biomechanical behaviour of the human hand, based on a space of
reduced dimensionality.
The present research project deals with the generation and
implementation of hand grasps of underactuated artificial hands
taking into account the characteristics of the human grasp, that is,
transferring to an artificial hand the knowledge that we already
have about the human grasp. Given the extent of this topic, the
project is restricted to the generation of the most common grasps in
activities of daily life: the cylindrical (power) and the pinch
(precision) grasps.
Firstly, the work proposes the development of a metric, indicator or
protocol for the evaluation of artificial hands that allows
determining its possibilities of control and emulation of usual
activities carried out by the human hand. For this purpose, a
theoretical comparative study as well as an experimental evaluation
of different hand designs is established as an objective. These
designs must be manufactured, sensed and controlled by low cost
systems.
Secondly, the development and implementation of intelligent control
algorithms of artificial hands is taclked, both from the point of
view of the low level control, having as objective the multivariable
control in real time of the different controllable degrees of
freedom of the system, as the high level control, generating grasps
as similar as possible to the human hand grasps.
Finally, the use of new and high-tech communication interfaces
between the user and the prosthesis is proposed, which imply
alternatives to the usual control by means of electromyography
signals. For all this, grasping experiments (cylindrical and pinch)
will be carried out with the artificial hands printed and will be
compared with those made by the human hand. This experimentation may
be of use to validate and refine the proposed metric in the first
objective of the project.
The proposed research can be of great interest in various
applications, among which can be cited: improvement of the design of
bionic or prosthetic arms, development of efficient robotic hands
for domestic or industrial applications or the development of haptic
virtual reality devices.
