Abstract: A time-efficient and low-cost path planning strategy is proposed by designing and using a fusion algorithm composed of an improved sparse A^* algorithm and a bio-inspired neural dynamics model, and it is an optimal strategy for a rotorcraft UAV (unmanned aerial vehicle) when performing non-collision flying tasks in three-dimensional low-altitude cluttered mountainous environments. The bio-inspired neural dynamics model is integrated into sparse A^* global optimal search to adjust local paths in order to speed up the formation of the final optimal path in the proposed fusion algorithm, and the neural dynamics model is adopted to obtain and process local dynamic information from the environment in real time. Therefore, the online path planning is realized by the fusion algorithm, and dynamic path planning problem is solved, which is impossible for the traditional best-first search algorithm. Experiments are carried out in an emulational 3D task space of multi-peak mountainous environment, especially for the concave mountainous environment. Experimental results show that the proposed fusion algorithm not only reduces complexity and time consumption of A^* algorithm, but also takes the cost of the path into account which isn't considered in the bio-inspired neural dynamics model. Furthermore, it can cope with unexpected threats in the task space on line. Then finally, a low-cost and high-quality path is planned out to reach target position safely and quickly for a rotorcraft UAV flying in cluttered environment containing both static and dynamic obstacles.