0
\$\begingroup\$

As depicted in the image, I'm looking for sensors to incorporate in a project where I have to detect object presence in short range (10-100mm), speed (passing from one sensor location to another, 0.1-10m/s), distance and orientation (just roughly). Space is fairly restricted, and I only have the option to detect object from one side. After a fair bit of research, it seems the only option I have is an emitter-and-receiver type of proximity sensor.

LIDARs draws too much power and only good in the long range. It might be an overkill for this purpose, and they also fall in a price range that I can't afford (USD 100+). Ultrasonic sensors are cheap but may fail detecting object distance if the orientation exceeds a certain angle, as shown in an experiment carried out in this video (VL53L0X vs HC-SR04 experiment).

VCSEL based Time-of-Flight sensor (VL53L1X ToF Distance Sensor) and IR Range Finder (Sharp GP2Y0A21YK0F IR Range Finder (Distance) Sensor) are affordable (USD 10-30) and accurate enough in short distance, and don't seem to be affected by color and orientation of the object. All is nice if the object only moves slow enough.

Detecting object presence at lower speed is not an issue, and so is reading its distance. Orientation can be calculated by changes in distance over time. But ToF samples only up to 50-60Hz, and IR Range Finder sensor has at update period of just 40ms (or 25Hz sampling rate). If the object is moving at its maximum speed of 10m/s (which is really not all that fast), a ToF sensor sampling at 50Hz (or 0.02s period) can only trigger a reading if the objects are at least 200mm apart.

Am I looking at the wrong sensors for my application? Or have I misinterpret the specs - as I would expect sensors taking advantage of the speed of light should operate at least some few orders faster than ultrasonic sensors.

enter image description here

\$\endgroup\$
4
  • \$\begingroup\$ why do you expect a sensor to measure speed and orientation? ... that should be the job of software ... the sensor needs to measure distance accurately, nothing else \$\endgroup\$ – jsotola Mar 10 at 5:12
  • \$\begingroup\$ Correct. They will done in the software but sampling rate of most proximity sensor I find can't detect the distance accurate enough like I showed in the drawing: an object moving 10m/s and the best ToF sensor I find only catch up to 50Hz. Perhaps my question was not worded properly \$\endgroup\$ – KMC Mar 10 at 5:18
  • \$\begingroup\$ perhaps a camera and CV could be used ... i do not know for certain, but something like Kinect could work \$\endgroup\$ – jsotola Mar 10 at 5:32
  • \$\begingroup\$ @jsotola, I like to avoid having computer or device running OS. Other than to save money, I envision the system can be driving by microcontroller or Arduino-like evaluation board and display simple information on a dot-matrix LCD. \$\endgroup\$ – KMC Mar 10 at 5:38
1
\$\begingroup\$

I have had good success using IRDA Rx and recessed emitters using a binary code to filter out interference up to 1m transmission and 1/3 m reflection at 100 kbps with 7 parallel sequential emitters and 7 detectors. The trick for my app was to have 1 deg beam width by choice of 6 deg emitter and recessed in a non-reflective hole to reduce aperture window further. .

But yours is a tricky optical problem that demands good eyesite with cross-eyed emitters at various angles to detect depth of an undefined object without getting distance walls reflecting more than the object and using signal threshold with AGC to capture the reflection fast enough to meet your acuity requirements.

Stereo Vision cameras or fast rotating mirrors such as those used by high value autonomous vehicles is the obvious choice, so I cannot give you any better solution than a plurality of BP narrow beam modulated carriers with a Bandwidth and fast AGC design to capture your objects using an array 1 deg apertures. We used a PIC CPU with 500 bytes of careful expert crafted code to detect moving objects with a signature shape which is different, and only 1m/s with parts as small as a resistor lead other as big as a shopping cart.

\$\endgroup\$
1
\$\begingroup\$

I'd say an adequate video system should be the preferred way, especially if your target has features to lock the detector on. Depending on the placement and possible path you could get away even without a stereo setup. Difficult to fit a reliable system in you budget however.

You could also try evaluating a millimetric wave radar, they are becoming cheaper since they are being massively used as anticollision in car.

I guess some experimentation is in order.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.