An intelligent robot for flexible quality inspection

Quality inspection is a task in every production concerned with a product's functionality, as well as its appearance. While dimensional or functional defects can easily be detected automatically, cosmetic defects are still found by skilled personnel only. For high-output productions companies cannot afford to inspect every part for its flawlessness, yet the ever growing demand for impeccable products asks for novel inspection solutions. We introduce an intelligent robot capable of performing visual quality inspection for geometric defects, but also for defects that fall short of any conventional description. A robot vision system is described that can intelligently analyze component's cosmetic quality along with an implementation scheme introducing the usage of the same. Depending on the object to be inspected and the level of detail, a constant processing time far below 1s can be achieved, allowing the implementation of an economic cosmetic inspection solution. Our intuitive inspection robot is reliable, highly adaptable, and fast, which can address cosmetic quality inspection utilizing an artificial neural network approach, yielding superior results where traditional template-based solutions lack functionality. An implemented case study is presented exhibiting an inspection success rate of 100% for two different scenarios.