A Pick and Place Mechanism with Vision to Sort Objects of Different Colour and Size

The traditional pick and place mechanisms in manufacturing and automation often lack adaptability, precision, and cost-effectiveness, when handling objects of different sizes and colours. This paper describes an innovative pick and place system that employs vision sensors to enhance the size and colour detection capabilities. The proposed system utilises a pre-trained monocular depth detection model along with a custom-made size detection algorithm that can measure object's dimensions accurately without having to resort to stereo vision systems. For colour detection, we adopted HSV colour space to ensure strong operation even under different lighting scenarios. A smart phone camera is used as a vision sensor to test the performance of the system. It is demonstrated that the proposed system greatly increases the pick and place machine's efficiency and versatility, thereby making it more appropriate for today’s fast-changing industrial settings.