Nytt AB was formed on the basis of a thesis aimed at simplifying the concepts of smart manufacturing, helping out small and medium-sized enterprises improve their manufacturing setup. Our mission is to make manufacturing more efficient, operator friendly and optimize the utilization of available resources. We strive to provide simple and easy to use solutions developed from readily available technology to make installations for customers hassle-free.

Background

Smart manufacturing is a buzzword that is trending in the field of engineering. The concept has helped companies evolve from the era of automation to cloud computing and connected machines. The technology enables an improvement in productivity, efficiency, and flexibility of manufacturing systems and large companies with considerable financial resources have developed solutions suited to their requirements but the solutions have largely remained unstandardized. This means that small and medium-sized enterprises that have limited resources do not have an opportunity to adopt smart manufacturing solutions and grow in an industrial setting. In an age where there is interdependency between companies, it is important that companies grow together and help each other for the betterment of the production ecosystem.

Technical description

CNC machines, in general, have a light indicator on them that defines whether they are running, waiting or stopped. These lights are similar to the traffic lights and are a clear indication of what state the machine is in. This light indicator is monitored using an android phone and the data is processed to trace how much the machines are actually working. The data is collected using the camera on the android phone by using a combination of motion detection to gather images for a data set to train a CNN model and using the YOLO detection engine to facilitate edge computing. The data from the state transitions in the machine gives an idea to the manager regarding the utilization of the machines and provides a clear idea of where there is a potential for improvement. For this thesis, our aim is to use improve the strategy used in the current system as well as use the other existing sensors on the phones like accelerometer, sound sensor, gyroscope, lux meter and so on to see how that data could be used to complement our existing solution and improve the quality of data that is being collected

Aim

• Strategy study of an existing CNN model with the YOLO detection system and suggesting improvements to reduce data set training time and improve detection quality.

• Analysis of how the accuracy of a detection engine with a limited amount of data sets be improved.

• Data analysis of sensors on mobile phones to improve the existing solution.

• Best fit scenario based on the location of the sensor, distance from the light indicator, distance from the cutting tool of the machine and other parameters.

• Combining sensor data from the phones and light data to improve existing data quality and find patterns related to a CNC machine.

• Calibration of sensor data to fit new installations.

• Investigate other existing sensors in the market that could simplify the solution even further