Electronic Design reader Colin Hartley asks:
"I am looking for a method of detecting a human body up to 50 feet away, and then I would like to determine his/her approximate x-y location within the detection area/zone. I thought of using two passive sensors, mounting them on two stepped motors, and scanning the area, but I can't seem to find narrow-beam sensors. Is there another method or product I can use to achieve the end result?"Answers:
Your human detection system could probably be done with either thermal or IR imaging. A few reference points in the field of view outlining the detection area should be sufficient to capture the relative position. Some custom imaging software could handle the trig required to do the triangulation calculations. The aforementioned reference posts could also serve to approximate a person’s size and weight. The frame rate could approximate a person’s direction of movement and speed. This is similar to traditional radar systems except that the imaging technique does not require irradiating your subject with RF.
Use standard sensor parts and put them at the end of a tube to reduce the viewing angle.... If reflections downt the tube produce too wide a viewing angle, put ring baffles with non reflective surfaces in the tube to reduce sidewall reflections. Keeping it simple.
My first thought is Colin could use a number of ultrasound transducers located at various points in the area and process the echoes to determine the persons location. You can buy the ultrasound devices from a number of places. (try http://www.roguerobotics.com/store/)
The pattern of the transducers and the shape of the area would determine the number of transducers and locations to make sure every part of the area is covered.
I did a quick trial with a distance measuring device from a local retailer and it found me at about 2m away (the length of the stick I used to press the button).
A few years ago a product was developed by me for a camera focusing aid, used the method of stereo vision by scanning mechanically, two small chip cameras (CMOS CCD) and processing the angles to achieve overlapping images with the focal point of the camera resolving an 'in focus plane'.
The inventor and some pictures of the device can be found at http://pictures.care2.com/view/2/492949080 This product won a Gemini Award in 2003/4
I am confident and often thought about using a similar system but replace the CCD's with Passive Infra Red (PIR) sensors and we could 'track' the x-y location of a person.
All this assumes there is only 1 person in the field.
If not, the PIRs would view the same scene via scanning mirrors which would then indicate multiple 'peaks' of IR sensed by the system. Integrating the two 'images' would then map the area and produce a typical location field of presence using triangulation.
My estimated budget for development of a prototype would be about $50 000.
40 years ago I designed some seismic systems for intrusion detection using geophones for oil exploration. Using a geophone array, you can triangulate the location of a single intruder. More than one presents problems. A single intruder can easily be identified as to location within a 50-foot radius. Activity outside the area of interest can be programmed to have no effect. HOWEVER, any seismic activity will cause noise that could give false readings. You can use ANY wide beam detector and triangulate. The RF/IR sensors present serious timing problems. Sound wins over sight.
Do you have any solutions to this design problem? Please e-mail your ideas to Lisa Maliniak, eMedia Editor.
Readers seeking help with their own design problems are encouraged to e-mail Lisa Maliniak, eMedia Editor, for posting in future editions of Electronic Design's Helpline.