Illumination Effect Evaluation Algorithm for Vision-based Robotic Inspection

Recently, the demand for robotic inspection of parts with complex shapes has been increasing. View planning is critical for robot inspection to determine the position and orientation of a vision sensor. Much research has focused on view planning that satisfies the sensor constraints, such as the FOV and view frustum. However, when an active illumination is used in the inspection system, such as a lighting device or a projector, image quality may be reduced due to the illumination effect caused by light reflected from an object. Thus, this study proposes an illumination effect evaluation algorithm for vision-based robotic inspection which generates a lighting model that considers global illumination at low computational cost and quantitatively evaluates the illumination effect on vision sensors. This method achieves the computational complexity of O(n) for generating a lighting model using the light intensity approximation model. Simulation results showed that the computational time of generating the illumination effect for 10,000 points is up to 1.32 s, and the illumination effect can be obtained quantitatively for various viewpoints by using the proposed method.