Teaching processing using a model of a workpiece is performed by accepting at least one input of a target value for a response time and a tolerance value for a detection error as a parameter representing a condition for a displacement sensor to operate. A CPU during teaching processing determines a maximum exposure time while it causes repeated detection processing by a light projecting unit and a light receiving unit, calculates variation in measurement data of an amount of displacement, and derives the number of pieces of data of moving average calculation suitable for a value for the input parameter as a result of operation processing using the maximum exposure time and variation in measurement data. This number of pieces of data is registered in a memory and used for moving average calculation in a normal operation mode.