Low Cost High Resolution Position Sensor with Sub-micron Accuracy

High accuracy position sensors are essential components of many precision machines. Currently available commercial sensors can achieve nanometre precision, but they are expensive because they require accurate machining and calibration. Such solutions include strain sensors, capacitive sensors, laser interferometers and linear encoders. A cost-effective sensor capable of sub-micron precision readings, besides being a competitive alternative to the existing solutions, can be used to add positioning feedback in inexpensive systems, improving machine performance and process quality. Therefore, this thesis proposes a technique to achieve sub-micron precision but using only low-cost components. The approach consists of imaging a strip composed of a sequence of bands and correlating the images taken at subsequent instants to determine the displacement. This thesis aims, first, at analysing the technique from the mathematical point of view, studying and optimizing the correlation computation. It then presents an implementation of the sensor written in C, which is executed on an ARM board running Linux and uses a USB microscope. Next, the performance of the implementation is compared on different devices and with various configurations. Eventually, the sensor is tested with the help of a 3D printer to verify the correctness of the readings.