Orbit Planner: Obstacle Uncertainty-Aware and Probabilistic Cost-Weighted Trajectory Planner Under Perception-Limited Constraint

Motion planning of UAVs is crucial for enabling autonomous flight, ensuring safety and efficiency in navigating complex environments. However, current methods lack consideration of obstacle uncertainty arising from perception-limited sensor ranges. Therefore, we propose Orbit Planner: an Obstacle unceRtainty-aware and proBabilistic cost-weighted Trajectory generation method under perception-limited constraint. As a result, the adaptability of UAVs to unknown environments is improved, enhancing the success rate of planning and ensuring safety. Specifically, a probabilistic cost-weighted kinodynamic search method is proposed, which computes trajectory costs using probabilistic weights. On the basis, different penalty weights are assigned for traveling in free and unknown spaces, ensuring that paths are prioritized to be found from free space. Additionally, an obstacle uncertainty-aware planning method is leveraged to actively avoid potential obstacles in unknown space, reducing the flight risk caused by obstacle uncertainty. Finally, comparative simulation and experimental validation demonstrate that the proposed algorithm can enhance flight safety and increase the replanning success rate.