A Coordinate Measuring Machine with Auto-Learning Capability

Coordinate Measuring Machines (CMMs) are widely used by auto-industry to measure the geometrical features of auto parts. For a CMM to accurately measure the geometrical features of a part, a model has to be developed and added to the CMM library. This process is time-consuming and may take several weeks to be completed. Moreover, vibration as an environmental factor can potentially distort sampled data and reduce measurement accuracy. This paper presents an auto-learning algorithm to reduce the time required to add a new part to a CMM library. Moreover, a cost-effective solution to reduce the effects of vibration on measurement results is presented. An improved Modified Multi-Class Support Vector Machines (iMMC-SVM) algorithm is developed to determine the geometrical features of parts through comparison with a reference part using a laser-based CMM. Experimental measurements are conducted using a prototype CMM design by the research team to validate the proposed solution. The results indicate that the proposed method reduces vibration noise by 6.18%. Such a noise reduction significantly improves the overall measurement precision.