Motion planning is a pivotal area of research for intelligent mobile robots, including unmanned aerial vehicles (UAVs). This paper focuses on the motion planning problem for fixed-wing UAVs and proposes a modified Dubins-Rapidly-Exploring Random Tree*(MD-RRT*) algorithm. The algorithm refines the sampling function by eliminating random angle sampling. It also enlarges the safety radius to reduce the computational load of collision detection. Then, a proof for the reasonable boundary value of the enlarged safety radius is provided. Additionally, it outlines the applicability conditions of the proposed algorithm compared to the original Dubins-RRT* algorithm. Through simulations, it demonstrates that the proposed algorithm enhances the overall computational efficiency of the algorithm at the expense of certain sampling boundaries and achieves shorter average paths.