To address the challenges of path planning for unmanned surface vehicles in complex waters, this paper proposes an improved ant colony optimization（ACO）algorithm based on uneven distributed pheromone and multi-objective optimization.

First, a probabilistic roadmap method (PRM) is used to generate an initial path. Based on the orientation information of the initial path and the endpoint, the ACO algorithm is guided to unevenly distribute the initial pheromone, resulting in higher pheromone concentration of the initial path and endpoint while decreasing the pheromone concentration of other grids in mapping according to the initial path-endpoint distance. Therefore, the problem of the ants' blindness in the preliminary path search improved, the calculation time is shortened thereof. Next, an objective function is constructed for solving the multi-objective path planning problem, and the weights are set to balance the relationship among the safety index, the energy consumption, the tortuosity, so as to providing diversified path to meet the requirement for different scenarios, moreover adaptively adjust the increment of pheromone to strengthen the influence of high-quality path in the whole ants colony based on the pros and cons of the planed paths. Meanwhile, to optimize efficiency improvement, an adaptive adjustment strategy of heuristic matrix coefficient is established, incorporating cosine modulation factors pertaining to iteration numbers. To obtain the global optimal path, quadratic optimization is carried out to reduce turns and turning amplitudes. Finally, on the basis of the maps of two real lakes—Lake Xiangdao (Huangshi ) and Lake Qiandao ( Hangzhou), the experiments are conducted to compare the effects of path planning using the proposed algorithm with that of other algorithms, i.e. traditional ACO, A* algorithm and improved ACO algorithm.

The results indicate that the proposed algorithm has the shortest planning paths, which is 61.71% shorter than that of the traditional ACO algorithm, the farthest distance from obstacles, and the smallest tortuosity. The running time of the algorithm is also improved.

The experimental results show that the proposed algorithm can reduce energy consumption during navigation, as well as the number of turns and turning amplitude, improving the smoothness and safety of the planned path.