You can do a simple integrated circuit with input's and output's pull behavior being pull down. It will convert error and floating to 0 and keep 1 and 0 as is.