Automatic Classification of Multiple Objects in Automotive Assembly Line

Abstract

In automotive manufacturing, assembly tasks depend on the correct identification and selection of workpieces, that may belong to more than one type of vehicle to be produced on the same manufacturing line. Usually, those tasks are conducted essentially by humans, which recently have been complemented by artificial perception provided by computer vision systems (CVSs). Despite their relevance, the accuracy of CVSs depend mostly on the environment control, providing appropriate lighting, enclosure and stops for images to be collected. This makes the solution expensive and overrides part of its benefits. This paper proposes a deep learning-based alternative to detect and classify multiple objects in automotive assembly line. Results show that the detection system has acceptable accuracy, it does not require any interventions on the production line, and it keeps its cycle time. The approach is illustrated by two examples of object detection over a real automotive assembly plant.