Ant intelligence for solving optimal path-covering problems with multi-objectives

Conventional methods have difficulties in forming optimal paths when raster data are used and multi-objectives are involved. This paper presents a new method of using ant colony optimization (ACO) for solving optimal pathcovering problems on unstructured raster surfaces. The novelty of this proposed ACO includes the incorporation of a couple of distinct features which are not present in classical ACO. A new component, the direction function, is used to represent the 'visibility' in the path exploration. This function is to guide an ant walking toward the final destination more efficiently. Moreover, a utility function is proposed to reflect the multi-objectives in planning applications. Experiments have shown that classical ACO cannot be used to solve this type of path optimization problems. The proposed ACO model can generate near optimal solutions by using hypothetical data in which the optimal solutions are known. This model can also find the near optimal solutions for the real data set with a good convergence rate. It can yield much higher utility values compared with other common conventional models.