Traditionally, the inspection and correction of the painted surfaces was carried out by workers under quite uncomfortable conditions of high luminosity level and with limited time. This may cause ocular fatigue and leading to less reliable defect detection.
The completely automated inspection system (inspection tunnel) for defect detection and monitoring of painted car-bodies based on computer vision techniques and distributed is composed of:
- The vision system based on cameras for image acquisition and lighting subsystem along with sophisticated defect detection and classification algorithms.
- The Cartesian mechanical structure for supporting cameras and moving lighting elements.
- The set of displays as man-machine interfaces, where defects are highlighted for later manual repair.
The image acquisition subsystem is composed of digital CCD high resolution cameras acquiring images at high frequency, assembled with high quality optics with variable focal length. These cameras are fast enough to meet the temporal requirements established for the factory. Each camera covers a part of the body where some overlapping is established for adjacent cameras in order to avoid missing defect detection.
The lighting subsystem uses high frequency fluorescent lamps mounted on several arcs moving at a constant or variable speed.
A Cartesian robot constitutes the inspection tunnel where the fixed structure supports the acquisition subsystem while the lighting subsystem is on the mobile structure (gantry type).
A display subsystem based on several screens, in the polishing area, provides information about the nature and location of the defects. Priority of repairmen is given by displaying large defects in red while small defects are displayed in green. These criteria are established by the quality control department.
The inspection tunnel has been running since December 2007 at the FORD Factory of Almussafes (Valencia, Spain), simultaneously detecting the defects of all of the produced cars (Focus, Fiesta, C-Max). The system is able to inspect more than 2,000 vehicles per day, in a processing time of less than 10sec.
With this system, it is possible to detect and classify defects with an improvement greater that 90% with respect to human inspection. Take into account that workers were unable to discover some defects due to their small size or their less-accessible location on the product.
In addition, the new vision system improves the working conditions and comfort levels of the workers on the factory paint lines.
As a result of all the research, an international patent (ref. PCT/ES2007/000236) has been generated based on artificial vision and processing algorithms, in addition to another international patent (ref. PCT/ IB2010/052193) describing the industrial system already implemented at the Factory.