And every time they unintentionally failed to second guess, what happens? A full police response unit? Revocation of license? What if their ankle is in the footwell of a car? Or in the middle of a house? Full police response unit?
How'd you work the batteries? Charge them with recharging their own as a license condition? Send someone round every 2-3 days?
The difficulties I see are that you say that you recognise that risk elimination is impossible, yet that's what you seem to want. There are, IMO, additional risks in aiming for risk elimination. Not least that its likely to involve exponential increases in resources for increasingly incremental gains. And in a context of finite resources, that means fewer resources dedicated to "low" and "medium" risk offenders. I've got the stats upstairs somewhere, but off the top of my head I think it's 80% of serious violent & sexual reoffending that is committed by people who aren't considered "high risk." Every bit of resourcing that's channeled into addressing the "highest risk" group who're responsible for the other 20% of serious recidivism is resourcing that is no longer being allocated to monitoring or managing other people, and other risk groups.
Risk elimination makes most sense in a context that presumes no context, few resource limitations, and no adverse consequences. There are probably ways of improving things, including monitoring - which would be awesome. But those kinda need to be real-world solutions that are likely to offer more positives / advantages / reductions in reoffending than the current situation. And which can take account of real-world behaviour, and what might be needed to respond to all the hundreds or thousands of potentially insignificant / harmless infractions that increasingly sensitive systems might create.