Highly articulated biomechatronic systems, such as humanoid
robots, exo-skeletons, and medical assistive devices, are gaining high
popularity in recent years because of their potential applications in
human-centered environment for health care, service, security, and entertainment
purposes. Since the mechanical structures of these systems posses very large
number of degrees of freedom, the motion generation and control of these
biomimetic systems are very complex. The common challenges are in re-producing
movements of their counterparts in the real world and in synthesizing new
coordinated movements. The motion produced by the human or the man-made systems
must obey physical laws and adapt to surrounding environments. For example, to
produce human-like behaviors in Honda Asimo robot, the researchers must first
record complete human motion sequences and generates the corresponding robot
movements through an off-line based learning model. This is a time-consuming
task due to the tedious procedure and equipment setup.
Hence, the main purpose of this project is to develop a
high performance un-tethered wearable human motion processing system for
real-time sensing and processing of anatomical motion data. Embedded sensing
techniques for human movement detection, embedded motion data fusion and
processing capability, ergonomics of the wearable system, interaction between
the wearable sensing system and the physical/virtual environment will be the
main research issues. Major applications of this human motion processing system
will be in the real-time control and manipulation of virtual figures in
interactive media, games and films, and articulated physical biomechatronic
systems, such as humanoid robots, exo-skeletons, and rehabilitation devices.
In this project, we propose to use optical fiber technology
to develop a new type of sensors in the strip form to detect bending and warping
of limbs and joints as the fundamental element of the body motion processing
system. Fiber optics enables the sensing device to be free from signal
interference, obstruction or shadow effect. The sensor strips can be woven into
a web directly worn by the human or embedded into the clothing. The patterns of
the strip sensor web can be configured for a specific application as well as
customized based on the biometric data of the user. An embedded processing unit
will handle the bending information of all strip sensors to produce the complete
body motion data based on the patterns of the strips and biometric data of the
user in real time. The real-time motion data can be used for control and
manipulation of physical and virtual systems through wireless communication
capability built into the system.
|
 |
 |
|
(a) Concept of SmartSuit |
(b) System architecture |
Main tasks include:
·
Prototype of a wearable embedded body motion processing
system in the form of a “smart suit” and a stand-along unit
·
Real-time
motion generation
method and APIs for biomimetic mechatronic systems
·
Prototype of
demonstration platform for the motion sensor suit
