Sensor de rango IR Sharp GP2Y0A02YK0F - 20 cm a 150 cm

Good product, robust and well made, works across the supply voltage range, one little niggle the product specification diagram, would be more helpful for people if the supply + and - where to be given a red and black colour name. Also, the pinout leads on the circuit board are long and there is a worry that they could touch any metal support bracket, that is the reason I suggested worth getting the support bracket, as I suspect it removes that worry (nice touch if it was to be supplied free!). It is worthwhile, buying the support bracket for this product as it would make mounting a lot easier. The range is accurate but is proportional to voltage.

It does what it says and I like that there are several versions for different distance ranges. The analog output is easier to deal with than reading a pulse width on a sonic ranger type sensor. Also, and this never occured to me beforehand but sonic rangers have a hard time dealing with clothing so if its peeps your ranging then infrared is the choice. OTOH since its analog you need some kind of interface for a raspberry pi.

This type of sensors from Sharp shares two issues: One is that their output is noisy, not quite random but occasional jump style noise. Median averaging can be helpful in this case but it limits the maximum bandwidth of the sensor. Second issue is more tricky. When the object is out of range, i.e. too near or too far then it produces unexpected value. For example object at 2m (out of range) produces the same output as object 1m (in range). When it comes to environmental effect, i.e. surface condition of the object or effect of day light, etc., it is more robust than other type of sensors such as ultrasonic type.

I have used Sharp IR range finders all the way back to the 90's They have always been a great reliable sensor. They are often used in commercial and production level robots. You have to understand the nature of how it works, to understand its limitations. It is simply sending out an IR blast than looks for a return, of any level, to gather how far the object is based on the amount of returned light. So the surface, shape, and size, of the object can change the results you get from the sensor. So like with any object tracking of this nature it is best to take the average of multiple sensors, and or multiple samples of a single sensor to get a better overall average, and remove anomalies. NOTE: These sensors work great in groups and do not interfere with each-other. With that said, the consistency across the sensors I have ordered over the years has always been the same, the notes mentioned above, are known, and apply to all sensors equally, so knowing this, you can allow for it in your application. For known variability is much better than pure random. For the most recent install, I picked up 8 of these for a leg detection system, and at 4-7 Feet I found this sensor was able to get reliable data that was very usable. For people may not often think of these as good 'human' detectors, but they have a very fast response, limited beam width and fast 'resets' Unlike other sensors like PIR, and SONAR. One tip I will give is get the harness, you will thank me later, when you try to source the proper connector and wire combination. The one available here by the Robot shop, is a perfect mate. I would recommend this sensor to those who have it in mind. You can not beat the price, even for just a test prototype.