Proximity Sensor (People Detector)

ɟɐsʇǝppʎ said:
Have you looked into using the XBox 360 Kinect, this could give you much greater control and then you can also control brigtness and anything else you want and im sure a few people would love to see someone do something for our hobby using this great device.

Here is a link to the Kinect SDK
http://www.microsoft.com/en-us/kinectforwindows/develop/
Click to expand...
And also the Kinect SDK only works for the "Kinect for Windows" hardware, which I believe is a bit more expensive than the one for Xbox 360. It's pretty much the same hardware, but the xbox hardware is subsidised by game sales, so the kinect SDK is hardcoded to not work with xbox kinect.
 
Here's the main PCB from the ultrasonic parking sensor kit I bought on eBay last week.

davidavd Ultrasonic Parking Sensor PCB from ebay.jpg

The four sensors (included in the kit) plug into it along with power and the display unit. The display unit has a small micro that drives the (very irritating) beeper and LED display. It would seen that the data is unidirectional - central PCB to display.

According to one web page I found, the wireless link (to the display) version uses a balanced data transmission method. This may explain why the a quick sniff with my com port tool did not seem to indicate regular async data. I'll hook up a logic analyser to it when time permits and decode the data format.

Anyway, once I get that sorted, I think these kits could make a cheap source of ultrasonic sensors and front end processing. Much easier and easier than sourcing all the parts, especially the little transformers.
 
The IR reflective sensor I got to try was an Optek OPB725A-18Z from RS Components. It was the only one I found that had a claimed range of more than 200mm (8") and didn't cost your first born!

R0414169-01.jpg


Some initial testing shows that (as I half expected) the claimed 600mm (24") range is really only achievable when sensing a reflective (to IR) object. A shiny new magazine cover did give fairly good results, although it had to be parallel to the sensor surface.

Range testing on people (ok, it was me in my black work pants) showed a usable range of only 200mm (8") or so.
 
David_AVD said:
Yes, but too expensive, too wide field of view, not robust, etc.
Click to expand...


Expense and especially robustness are issues (especially their combination - I would hate for a rain to destroy something that expensive!).


But the "width of field" should be excellent; just think of it as scores of very narrow range sensors in one package.
 
Some people have used low power lasers and cheap ccd cameras to detect presence and distance (via parallax - face both in the same direction a few cm apart - which pixels are illuminated depends on distance). Easy to get reasonable resolution in the range you want. The laser can be pulsed occassionally, doesn't have to be on full time (look for a region of the image which is consistently brighter when the laser is on).

I believe there are some assembled or kit versions of this in the robotics DIY arena.

Junior: I just saw a firefly on my jacket!
Mom: I don't see anything, but look at the way those lights have gone crazy!
Junior: there is is again!

I mention this because it has a very narrow window.


I think the quad ultrasonic parking range detector has great potential, if the sensor window width is OK, and if you can find some way to tap into it.
 
David, is that the data from the sensor electronics to the display head?

It looks like there's a consistent falling edge, but the rising edge can be about 1/3 or 2/3 of the cycle.

Looks like around 32-33 bits. Is this the full (repeating) pattern or only a fragment?

Does the pattern vary as the ultrasonic transducers get reflections?

(Or am I way off and this is what's going to the transducers or something?)
 
Yes, that's the data from the controller to the display unit. The bit patterns do change with sensor changes. I just haven't had time to look at it properly yet.
 
Have a look at this:
http://www.mp3car.com/hardware-development/121995-custom-usb-parking-sensor-interface-formally-pdc-usb.html

and this:
http://ishankarve.blogspot.com/2011/06/decoding-ultrasonic-parking-sensor.html

It looks like the unit *might* give 4 eight-bit ranges in centimeters, with a simple to decode protocol. I couldn't ask more much more than that! Cheap, 4 outdoor sensors, range in cm up to 2.5m, possible option to receive via wireless (stealing rcvr from display unit). Pretty great!

Color me officially excited by this - it could be just the interactive element I've been (lazily) looking for.

However, your diagram doesn't look quite the same. You have a short/long vs long/short (high/low), rather than short/short vs long/long.

Which one did you get from eBay, David?
 
Yes, I saw that project link a couple of weeks ago. Seems like the one I got is the wired version of that one.

I will know more once I get some free time to use the logic analyser on it again.
 
I've now analysed and documented a bit more of the serial data stream that the main board sends to the display unit.

The sensor data is in blocks of 8 bits (32 bits for 4 sensors) and I've worked out the relationship with displayed distance. The start bits are a bit weird however and I'm mulling over coding options.

Maybe I'll use a small PIC to translate the sensor data into standard async format (say 9600 8N1) for proportional readings. I could also do a version that just sets a pin (per sensor) high or low upon any detection distance.
 
You must log in or register to reply here.
Back
Top