Hello guys, I am making a device which detect the gesture from the user. I used Arduino Uno as the development kit and a K-band micro wave radar sensor. But I am not sure if the Arduino Uno support K-band micro wave radar sensor, if not, which development kit should I choose?
I want to use IPM 165 made by B+B sensors.
datasheet: http://www.innosent.de/fileadmin/media/dokumente/DATASHEETS_2016/Datenblatt_IPM-165_V8.5.pdf
The output is AC and under 2 Volts (peak to peak). Because the signal goes negative you should not connect it directly. If you offset the offset the voltage by about +2.5V the signal should be in a valid range. Then you can then use an analog input to get a raw signal. I don't know how you plan to interpret the signal.
Does it means that Arduino can only get the positive signal from the sensors? How should I apply the offset voltage signal to 2.5 volts? Thanks. 
It seems to me, that this type of sensor can't be used for gesture recognition, first page of the data sheet marks only two green squares
x- Movement
x- Velocity
Direction
Presence
Distance
Angle
probably, sensor only could tell if there is movement in front of it, w/o any info on which way the object is moving. Same way as regular PIR sensor would do
Long time ago since I played with radar for burgler alarms.
AFAIK radar type sensors only detect objects moving towards and away from the sensor (doppler effect).
My guess is that the analogue output polarity depends on distance variations.
PIR sensors detect sideways movements. When the (warm) object moves to another facet in the fresnel lens.
The datasheet has no info about sensor output impedance.
As johnwasser suggested, use a 1:1 voltage divider and inject sensor signal onto that.
Try two ~22k resistors, one from 5volt to A0 and one from A0 to ground.
And a ~47uF cap between sensor output and A0, with cap positive to A0.
Analog readings will be ~512 without, and higher or lower (direction info) with movement.
Leo..
I was trying so do something similar to project Soli developed by Google. (Home: Soli)
I think they are using the similar type of radar sensor. If not, what kind of radar should i use? IR sensor or some motion sensor cannot recognize hand gesture precisely and may easily affected by the environment (Temperature and lights). Using micro wave radar sensor with short wavelength is the only solution that I know.
Thanks for the circuit suggestion.
First off all, they build sensor in another ISM band:
"The Soli sensor is a fully integrated, low-power radar operating in the 60-GHz ISM band. "
Internal structure their sensor must include phase array antenna, and more important the processing of reflected waves. Probably, patented.
I did once gesture recognition project with HC-SR04 ultrasonic range finders, using 5 of them and arduino Due. Another option for you is https://www.sparkfun.com/products/12787
IR based.
I do not agree with your statement
netscapt:
IR sensor or some motion sensor cannot recognize hand gesture precisely and may easily affected by the environment (Temperature and lights).
Combining 4+ and more receivers, and measuring the slightest difference but not absolute level of signal, may amaze you
MasterT:
First off all, they build sensor in another ISM band:
"The Soli sensor is a fully integrated, low-power radar operating in the 60-GHz ISM band. "
Internal structure their sensor must include phase array antenna, and more important the processing of reflected waves. Probably, patented.
I did once gesture recognition project with HC-SR04 ultrasonic range finders, using 5 of them and arduino Due.
Thanks for your suggestion, but I am confused now. What is the major different of using sensors in different band? Why can HC-SR04 do the job but not the sensor I said before? I tried to find so article about this but failed to find one, is there any article you suggest?
Higher frequency 60 GHz vs 24 means lower wave length, and consequently better spacial resolution. Ultrasonic wavelength is about same size, btw.
It's not clear what is your experience in electronics to give any advice.
If objective of your project is to track fingers movement, than DSP is your research area. For hands tracking, have a look on this project http://www.instructables.com/id/Arduino-Solar-Tracker/ it gives a basic idea.
The signal from the radar sensor is the raw doppler signal: the sum and difference you get when you mix the outgoing carrier and the incoming signal that has been modified by reflecting off a moving objects at various distances. It may be possible to detect gestures. I would try putting the signal through a Fourier or Hartley transform to convert from the time domain to the frequency domain and then maybe use a neural net to classify the signals. I don't know how big an Arduino you would need to do all that.
The link to the Google Soli project page indicates that their radar provides both Doppler and Range information. The unit linked in the original post is Doppler only. Some of the prototype photos suggest a multiple channel receiver (1 transmit + 3 receive?) which would allow some level of beam steering potentially giving them azimuth and elevation as well as range positioning. It's not clear whether that's in the end product or not. While most of the magic is probably in the processing layer, what can be done there is limited by the amount of information available from the radar sensor.
I found an article about neural network on IEEE library, this gave me hope on using the dolpper radar to do my project.
Its worth a try before I decide to use other sensors. Again, thanks all you guys answer my questions and it helps a lot.