Abstract: The high-precision data has the characteristics of irregular lane lines and segmented measurement, leading to problems such as bending and twisting in the connection of main roads. To solve this problem, an algorithm based on slime mold optimization is proposed to achieve optimal serial fusion of main roads. First, a solution space for serial concatenation of main roads is proposed based on feature vectors; then, according to the principle that main roads should not exceed lane lines as much as possible, considering the length of each section of the center line of lanes, and making the overall main road as close as possible to the center lines of all lanes, an evaluation function based on key nodes of lane lines and center lines of lanes is proposed; next, solutions are proposed for the hybridization and direction outlier problems of lane center lines with intersections and lane center lines; finally, combining slime mold optimization, detailed flow of the algorithm is given, and in order to accelerate convergence efficiency, a neighborhood solution strategy is introduced into the stage of slime mold food capture. Experimental results show that the proposed algorithm shows good robustness in various serial fusion scenarios; introducing neighborhood solution strategy can improve convergence efficiency; compared with other optimization algorithms, it has higher search accuracy and convergence speed.

Key words: slime mold optimization, feature vectors, evaluation function, neighborhood solution strategy, direction deviation