Precision we can expect of an ultrasound-based localisation system

Question

I'm considering building an absolute, indoor robot-positioning system based on ultrasound Time Of Flight. Transducers will be ordinary, narrow-band, 40 kHz ones.

Based on your experience, what is the best exactitude and precision we can achieve with such a system?

I'm aware that the answer to such a question will depend on many factors, both hardware and software (if applicable), but I'm not asking about the performance of one solution or another, but about the intrinsic limitations of the ultrasound technology.

Answer 1

It closely depends on the method you use to determine location. That is, will you use lateration or angulation? If you wish to use angulation then you need to synchronize the emitters (or receivers, depending on which of the two is being localized), find the phase difference using correlation of signals and then based on that estimate the location. At this point, the most difficult thing is the synchronization part and that's what will affect the error the most. If you're using lateration (measuring the time of flight), you need to somehow know when the signal was generated and when it arrived at the target. For this you could use some RF transmission technology (ZigBee or WiFi or something). If your detection resolution is high enough you can get a pretty good accuracy (if your sampling frequency is for example 500kHz for example it means you can detect the minimal distance of around 0.6mm because that's how much the sound travels between the two measurements) but the thing to also consider is the latency in the RF communication. That latency can probably be measured and estimated to a degree but it will affect your precision.

All in all to finally answer your actual question, if all is calibrated well you can get an error bellow 1%. Check this out for example http://link.springer.com/chapter/10.1007%2F978-1-4471-2386-6_105 their results say 0.1% error.

Regards