We present a wearable, motion capture system
that can acquire emotions of an individual
almost anywhere over extended periods of time.
Our system consists of an array of small, low-cost, low-power
sensors placed on the body of a person.
The system is completely autonomous.
It does not require any external devices or signal sources.
In this prototype, we use eight signal sources and 18 signal
detectors as our sensors, our sources
are small, ultrasonic transducers
that sequentially emit pulses about 16 times a second each.
We use small microphones as signal detectors
for the ultrasonic pulses.
The detectors also contain three-axis accelerometers
and three-axis gyroscopes that sample accelerations
and angular velocities about 140 times per second,
significantly improving the precision and the frame rate
of the ultrasonic measurements.
Using the onboard microprocessor,
the accelerometer and gyro values
are digitally encoded and combined
with the analog audio signal from the microphones.
All signal sources and detectors are wired to the driver box.
The driver box consists of the following components--
18 second-stage amplifiers that further amplify the signal
from each of the sensors, 3 six-channel, analog-to-digital
converters that sample the signal from 18 sensors
at 150 kilosamples per second, a microcontroller that
synchronizes all HD converters and generates pulse
signals for the ultrasonic transducers.
The signal from the microcontroller
to the transducers is then amplified
by the transducer drivers.
The driver box is powered by a small, rechargeable battery
pack that lasts for more than four hours.
The data from all A-to-D converters
is transmitted using a USB 2.0 connection
to a laptop which records this data directly
to its hard drive.
The driver box and laptop can be carried in a small backpack.
Our prototype system was built from off-the-shelf components
for less than $3,000.
Here is a comparison between our reconstruction
and that of the Vicon optical system.
In this sequence, distance measurements
prevent drift in the left shoulder.
The mobility of our system allows
us to capture motion virtually anywhere, for example,
while driving a car.
We were able to track all limbs as well as recover more
detailed and subtle motions.
Even motions that are traditionally hard to acquire
are captured with ease using our system.