pir is a great idea if you can find one that will work in your environment. I experimented with couple of different pir sensors and abandoned the idea completely for all projects as they just trigger to often on too many "non triggerable" events.
IR might be made sensitive enough (tx powerful enough) to get 1-2m without too much troubles (add few leds in parallel and drive them with max rated current, you will get much more then 2m, much much more)
ultrasound is great thing, I tried to play with it some time ago but wasn't able to finish the project as pressure on job increased so free time got vaporised .. but you can find some nice ultrasound schematics on this forum and some nice pointers from few other "really smart ppl"
In any case, doing it with ultrasound will take
- TX capsule (not cheep)
- RX capsule (not cheep)
- demodulator circuit (amplifier, demodulator, all analog - not my area really )
with compared to IR one where you need few transistors to drive led's and one ir receiver that does everything for you (amplification, demodulation) the price for IR one is ~10x lower then for the US one .... also, IR is much easier to focus the sound