Case Studies



This system was developed for an Automobile company for Quality Assurance checks of ‘Gears’, a critical component of their automobiles.
Automobile parts can have surface defects like scratches or dents occur during the process of Manufacturing and cross-border transportation. This may affect the service life of the vehicle itself.
The inspection process is mainly performed by human vision, which is unstable and insufficient. The combination of artificial intelligence, automation and machine vision can do a far more accurate and faster job. However, it is a challenge to inspect such defects in a computer system because of imbalanced illumination, specular highlight reflection, various reflection modes and limited number of defect features.
Bevel Gears
Surface defect detection and dimension measurement of automotive bevel gears by manual inspection was costly, inefficient, low speed and low accuracy. In order to solve these problems, a bevel gear quality inspection system based on multi-camera vision technology was developed. The system can detect surface defects and measure gear dimensions too.
The system can detect knock damage, cracks, scratches, dents, gibbosity or repeated cutting of the spline etc. The smallest detectable defect is 0.25 mm and the precision of dimension measurement is about 10 to 100 µm. (depending on quality of cameras used) One inspection process takes no more than 1.5 to 6 seconds (depending on number of faults, size and the quality of images). Both precision and speed meet the requirements of real-time inspection in bevel gear production.
Top Surface
Bottom Surface
Requirement Analysis
The requirements of defect detection and measurement are as follows:
  1. Defect detection of tooth end surface, spherical surface and mounting surface, including knock damage, cracks, insufficient filling, folding, scratches, dents or gibbosity, repeated spline broaching, etc. Defects greater than ___ mm × ___ mm should be detected.
  2. Measurement of bevel gear's height, diameters of two circles, spline's big and small diameters, end rabbet's diameter. Measurement accuracy is ___–___ μm.
  3. Defect detection speed is about ___ s for each surface.
  1. The first level inspection is conducted to detect faults on the gear. If any fault is detected, then the rejection pneumatic pushes the gear to the rejection plate. If no fault is found, then it proceeds to the second level of inspection.
  2. The second level of inspection is done to check if the dimensions of the gear matches the specified dimension. If the dimensions do not match the specification, then the gear is pushed out of the conveyor belt by the rejection pneumatic, otherwise the gear moves on to the third level of inspection.
  3. The final level of inspection is done to check the dimension of the centre broach of the gear.
The multi-camera automobile bevel gear quality inspection system presented in this case study adopts advanced machine vision technology. It solves the problem of difficult inspection of complicated gears at high speed and with high precision. The research shows that machine vision can substitute most manual work in bevel gear inspection, improving the efficiency of production and the degree of automation.