Monitoring of robot trajectory deviation based on multimodal fusion perception in WAAM process

During wire arc additive manufacturing (WAAM) process, the movement trajectory of robot directly determines cladding layer position, therefore, monitoring of robot trajectory is particularly important. This paper proposes an approach for monitoring robot trajectory deviation based on multimodal fusion perception. First, a visual sensing system is constructed using monochrome camera and infrared camera, which collected visual image of molten pool and measured the temperature of workpiece sidewall. Second, the weld pool contour is extracted based on deep learning, and the histogram of oriented gradient (HOG) algorithm on the basis of trilinear interpolation is adopted to extract temperature distribution characteristics of workpiece sidewall. Finally, a prediction model for robot trajectory deviation is developed using artificial neural network. The experimental results indicate that the proposed approach for monitoring robot trajectory deviation has high detection preci-sion and strong generalization capability, it provides an effective means to ensure weldment quality in WAAM.