Page 145 - Control Theory in Biomedical Engineering

130 Control theory in biomedical engineering

and sensitive to many disturbances. They can be affected by abrupt changing
of the electrode positions, changes in electrophysiology due to sweat,
changes in the impedance of the electrode, grease, muscle layers and tissue,
and time (Artemiadis and Kyriakopoulos, 2008, 2011; Waris et al., 2019). All
these conditions may lead to inaccurate identification of user intent, chal-
lenging the reliability of the control system (Parker et al., 2006; Hargrove
et al., 2007) and necessitating some signal conditioning or modeling.
Despite the focus on EMG-based control, handwriting, which is an
important means of communication, has received less attention from the
assistive device point of view. Writing has inspired many researchers to pro-
pose models characterizing this biological process. Indeed, handwriting is
considered a mean of communication unavoidable for academic, profes-
sional and social integration (Defazio et al., 2010). The production of
graphic traces is a physical manifestation of a complex cognitive process.
It contains a lot of information that can characterize a person and express
the level of academic and social education and even the psychological state
(angry, relaxed, etc.) of the writer and their temperamental tendencies
(Viviani and Terzuolo, 1983). Writing is expressed through motions of
the upper limb, and with the availability of advanced EMG recording sys-
tems EMG-driven models can allow reconstruction of individual
handwriting.
Considering different kinds of handwriting, letters, geometric forms, and
even numbers, this chapter reviews the most effective handwriting models
using muscle activities of the upper limb. The considered structures are based
on input/output relationships between pen-tip coordination moving on a 2D
plane and EMG signals of the hand and/or forearm. We describe the different
handwriting approaches and present the advantages and the drawbacks of each
one to finally develop a summary table of the different presented black-box
models. The chapter focuses on both simple movements, such as drawing
numbers and geometric shapes in a single orientation, and complex move-
ments generated in combined directions like cursive Arabic letters.
The chapter is organized as follows. After a brief history of handwriting
models, we present a Kalman filter (KF)-based approach by Okorokova et al.
to mimic some numbers. In the next section, we develop the approach
presented by Zhang and Kamavuako. Then, in order to characterize
handwriting cursive letters and complex geometric forms from two EMG
forearm muscles, we present two models. The first is based on the velocity
of writing developed by Murata et al. and the second is a robust interval
observer model.