Added in some 100µF ceramic caps across each of the input/output lines that fun from the Arduino board out to garage components. This includes the lines that go out to the reed switches which detect the garage is open or closed as well as the lines that go to the garage door motor telling it to switch state. What I think was going on was that sometimes the spark of the garage door motor starting caused enough RFI coupling into those fairly long wires that it would cause a voltage spike and essentially render the Arduino into a hung state. Seems to be much more stable now for months at a time (instead of a week or so). YMMV on this one :-)