Taking the gear assembly of the reducer assembly line as a research background, the holefinding strategy for robot assembly with keyed circular pegs is proposed to solve the problem of low holefinding efficiency and success rate. in the circular hole-finding task, deviation domains were divided based on a static mechanism, position vector trajectory optimization in the step distance and direction was designed, and a deviation domain-force mapping relationship was established using a genetic algorithm-support vector machine (GA-SVM) classification algorithm; the accuracy of this algorithm is approximately 90.00%. Thirty groups completed the circular hole-finding task in an average time of 5.4 s. For the square hole-finding task, dual monocular cameras were integrated to identify corner points of the flat key and keyway. Image semantic segmentation based on deep learning was used for the corner-point recognition of the flat key to suppress the effect of changes in light intensity; the recognition has an average error of 0.39 mm. Coarse and fine adjustment circumferential deflection strategies were adopted sequentially. The 30 groups of square holefinding tasks exhibited a 96.70% success rate in an average time of 10.2 s. The proposed hole-finding strategy improves the efficiency and success rate of the gear assembly.INDEX TERMS keyed circular peg; learning-based robot hole finding; support vector machine; image semantic segmentation based on deep learning; genetic algorithm
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.