Machine vision driven Industry4.0 solution for sustainable foundry / manufacturing operation

Purpose and goal

The purpose of the project was to develop a camera-based hardware platform to be used in modern manufacturing industries, in e.g. India. One application is the measurement of manufacturing quality when producing molds in sand. A mold is created, in a machine, by compressed sand which is subsequently inspected visually. A camera prototype, adapted for the machine, has been developed to replace manual visual inspection. Due to Covid 19, the project´s focus changed to further development of software for 3D sensors such as stereo cameras.

Expected results and effects

A compact camera prototype has been developed. The camera is built into a robust chassis to withstand a tough industrial environment. The prototype is based on mature sub-components and it should work well in an industrial environment. Thus, small changes should be required to create a camera product that can be mass-produced. Our work with stereo cameras confirms that good cameras, larger distances between cameras and a powerful processor are required to be able to detect small obstacles at 5-10m. The solutions available on the market today do not meet these eligibility requirements.

Planned approach and implementation

Development of a compact camera prototype was carried out after review of user requirements and discussion with our partner in India. The camera is based on mature sub-components to minimize cost and development time. We have not received any feedback on how well it works in an industrial environment due to Covid-19. Software development, to support stereo cameras, has been performed in OpenCV for Python. A lab setup with industrial cameras, calibration and test targets has been used. Calibration of the setup, and a powerful PC are important to achieve a good performance.