transform of each signal from the two PIR sensors and
a frequency analysis can resolve the problem of false
detection. Figure 2 shows a time and frequency spectrum
of two PIR sensors detecting a walking human. Low-cost
circuit based on ADC and embedded electronics enables
prevention of false detection.
With the same circuit, time-frequency domain analysis
– thanks to a wavelet transform – gives more information
about the signal. It is possible to have a “signature” for
different kinds of motion.
You can see the result of the wavelet transform in
a scalogram view. The shape of the spectrum in the
time-frequency domain varies depending on the type of
motion. Figure 3 offers three examples: human motion
(a), pet motion (b), and human rolling on the floor (c).
Visually, it is easy to see one or many clusters in
the scalogram, depending on the type of motion.
From wavelet transform data, you can easily apply a
classification algorithm and detect different kinds of
movement very efficiently.
You can use the same wavelet transform-based circuit
for other types of detection such as:
• Human adult walking.
• Human child walking.
Properly applied machine learning or artificial
intelligence will open the door for new types of
intelligent motion detectors. These motion detectors
will be able to adapt to different environments and
detect more than just human motion. Using two PIR
sensors and a wavelet transform based on embedded
and low-cost electronics has many applications and
can solve many customer problems. ECN
Figure 3: Continuous wavelet transform of two PIR sensor
signals for walking human (a), walking pet (b), and human
rolling on the floor (c). (Source: Texas Instruments)
OPTIMAL CONTROL OVER
Transfer heat from one location to
another at the component level.