Adaptive Visual Inspection

The robot vision system A prototype robotized vision system has been designed, using a six degrees of freedom anthropomorphic arm, for self-adaptive vision based quality control on an assembly line for washing machines (WM). The robot allows displacing the camera in the workspace so to inspect a washing unit (WU) or a complete WM on a pallet. The vision system is equipped with a 1360x1024 pixels GigE camera and a RGB nearly axial ring illuminator. The illuminator can be controlled in colour and intensity. During an inspection sequence, the robot moves at a number of selected fixed positions and acquires a set of images (for example the series reported in figure 2); it therefore substitutes a fixed constellation of cameras. Positions can be reprogrammed depending on the model of the part to be inspected, allowing flexibility. Trajectory has been optimized by genetic algorithms. This implementation is realized for the inspection of the WU or of the complete WM and controls take place at different locations on an assembly line. Figure 2 represents a sequence of control to be done on a WU: three clamps and one electric connector, whose presence and position has to be verified. Similar sequences are collected depending on the inspections to be done. The robot vision system runs feature based matching algorithms to recognize the parts of the WU or WM under inspection. The robot vision system prototype Grace project The EU FP7 Grace project aims at integrating process and quality control within a production line. This goal is fully in line with the trend to develop modular, intelligent and distributed manufacturing control systems. The system is based on a collaborative Multi-Agent System (MAS) which operates at all stages of a production line and it is complemented by self-adaptive control schemes developed at the level of process resources and quality control stations as well as at line or factory level. The MAS aims to individually tune parameters of each product taking into account information collected during the whole production process, so to compensate production process variance. The innovation is the new vision of the production process which leads to a deep integration of process control with quality control and finally product value.