<p>Considering the uncertain small movement of obstacles, the new academic problem of how to generate the optimal trajectory such as no-collision, minimum energy and minimum time is a planner optimization problem. However, multiple solutions and a non-convex objective function make this problem difficult. Current trajectory optimization techniques exhibit certain limitations, as each method’s specified condition yields only a singular solution, and current methods struggle to accommodate novel scenarios. we present a two-stage optimal trajectory generating technique in this paper. In the first stage, a series of multiple trajectories corresponding to various patterns of the non-convex cost function of collision avoidance, smzongjie oothness are generated. To simplify the issue, it is converted into a dstinct problem where the density resulting from a distribution is estimated from a cost function. During the second stage, the safest trajectory is chosen based on the micro-motion situation of the obstacle. By satisfy the requirements of collision avoidance and smoothness the suggested framework not only meets the criteria of collision prevention but also achieves risk minization, which is helpful in selecting a better solution from many candidate trajectories. We assessed the presented approach by simulating specific tasks in a scenarios on a scanning robot.</p>