Machine vision simplified with distance and ranging sensors
thesisposted on 28.03.2022, 19:44 by Ryan George Barnes
Machine vision in industry is dominated by 2D cameras. These camera systems are very effective for object tracking, but are labour and skill intensive to implement and require powerful standalone controllers to process the images. 3D and 2D ranging distance sensors provide three-dimensional data which could easily be utilised to perform the same tasks, without the issues of ambient light changes which cripple a 2D camera's ability to function. Currently distance and ranging sensors are marginalised to quality control applications in the machine vision field, focusing on product fill completeness and profile consistency checking. These distance-based sensors have the potential to perform tasks currently done by 2D cameras in industrial vision application, but in a fundamentally different way. The sensors provide an enhanced way of looking at a scene that would be very useful in applications which require identification of shapes and object tracking. Implemented correctly, the ranging and distance sensors should provide better and more flexible performance. The ability for these sensors to detect the actual size and distance of these objects eliminates the need for estimating size and position, which currently requires an experienced programmer to teach the system what it is incapable of learning itself. Furthermore, if programmed correctly, a distance-sensor-based tracking system could be made far easier to implement than traditional cameras and therefore cheaper on labour and more accessible to less experienced users and companies. This thesis project will quantitatively compare the accuracy of multiple industrial distance and ranging sensors on a range of commercial consumer goods. It will also develop a program to automate the tests and feature a quick setup program to explore the viability of end-user self-installation.