近日，研究团队硕士生李天奕作为第一作者，张雪波教授作为通讯作者的论文《STUNS-Planner: A Spatiotemporal Motion Planner with Unbending and Consistency Awareness for Quadrotors in Unknown Environments》被机器人领域国际期刊Journal of Intelligent & Robotic Systems录用。

Abstract

        Motion planning is fundamental for the autonomous navigation of quadrotors in unknown environments, which often consists of path searching and trajectory generation. However, existing path searching approaches overlook the influence of the tortuosity of the path, and trajectory generation methods usually ignore optimizing the time allocation. They both threaten the security and limit the efficiency of autonomous navigation. In this paper, we propose a SpatioTemporal motion planner with UNbending and conSistent awareness (STUNS-Planner) to generate a safe, feasible and efficient trajectory by reducing the tortuosity of the path and improving time allocation of the trajectory. At the path searching stage, the unbending-aware path searching algorithm is proposed to search a local safe and unbending path before the trajectory generation. At the trajectory generation, we propose a fast scaling time adjustment method that can rapidly improve the efficiency of the trajectory while ensuring feasibility. In addition, to ensure the consistency between sequential replanning trajectories, a consistency-aware trajectory preserving and connecting mechanism is designed. The planner is integrated into a fully autonomous navigation system on a quadrotor. Benchmark com-parisons and real-world experiments verify the performance of our system in terms of safety, tortuosity (about 8%-13% lesser) and trajectory time (about 3%-8% shorter) compared with existing approaches.