Actually, it's not so simple. The only way to make it simple is to have the paper ballots be the official medium; electronic counts may exist as well, but in case of discrepancy the paper ballot rules. To avoid deliberate introduction of discrepancies, then, paper ballots must either be entirely controlled by the voting equipment (perhaps viewed through a window) or a machine must read the paper ballots when they're physically inserted.
There's nothing inherently wrong with viewing-window-based systems, but the mechanics can be a bit tricky especially if there's a requirement (as there should be) to allow voters to inspect their printed ballot and reject it (getting a new one) if it's incorrect. In that latter case, there must be a mechanical means by which the rejected ballot can be seen to go somewhere other than the ballot box.
Frankly, I don't see any advantage of using such a system in preference to an optical-scan balloting system. Disabled voters can be accommodated in optical-scan systems via the use of voting terminals which will mark optical-scan ballots according to their selections. Their ballots could then be counted the same as any others.
Given that one optical scan reader can service a substantial number of polling stations, I see no reason for governments to spend the extra money required for captive-ballot systems which would seem to provide little if any real extra benefit.