This is a solve-able problem. First, we need to gather data regarding the frequency of leaving these notes. I suspect this is practically exceptional given the

This is a solve-able problem. First, we need to gather data regarding the frequency of leaving these notes. I suspect this is practically exceptional given the required costs associated with requiring a signature (only anecdotal but even more expensive purchases I've made almost never need a signature). Once you collect data that's your baseline for the printer's reliability. It must work at a minimum that many times per delivery route (in practice it probably doesn't). There are dozens of options for small, portable printers that are relatively inexpensive. So even before you start custom designing a more reliable printer get the costs of purchasing them in bulk and pilot them in an area attached to mail carrier bags. Fading / inking coming off would be an edge case; the vast majority of these notes are going to be basically warnings that you need to pick up your package or asking them to redeliver it and you have a very limited amount of time to do this before it gets sent back. Packages have barcodes, have the printer scan that to instantly generate an appropriate label. It's not the easiest thing in the world but I could put together a pilot for this inside of a week with proper USPS API access. Get into a fast iteration cycle, increase reliability of printers and prints, automate changing of ink and paper upon delivery route completion through some docking mechanism. I almost want to just start building this thing, ha!