This paper proposes an exploration strategy in unknown environments for a team of quadrotor Unmanned Aerial Vehicles (UAVs). Based on the frontier information, the proposed strategy builds a roadmap of the explored area in form of a Sensor-based Selective Graph (SSG) using simple data trees of the frontier and the hub node only. In particular, the frontier data tree is utilized to consider the adjacent frontier sectors as one frontier sector, and the next target node is generated maximizing the coverage of frontiers at each movement of quadrotors. In addition, to expand the proposed strategy to the three dimensional (3D) workspace with quadrotors, a Multiple Flight Levels (MFL) approach is proposed to increase the efficiency of the exploration. Moreover, when a quadrotor reaches a dead end where no frontier exists, the efficient backtracking algorithm chooses the best path to backtrack efficiently with a graph map provided by the SSG. With these contributions, we successfully develop the frontier-based exploration strategy for multiple quadrotors, and performance of the overall approach is validated by numerical simulations and experiments.

本文提出了一个四旋翼无人机（UAV）团队在未知环境中的探索策略。基于前沿信息，所提出的策略仅使用前沿和枢纽节点的简单数据树，以基于传感器的选择图 (SSG) 的形式构建探索区域的路线图。特别地，利用边界数据树将相邻的边界扇区视为一个边界扇区，并生成下一个目标节点，以在四旋翼飞行器的每次运动中最大化边界的覆盖范围。此外，为了将所提出的策略扩展到带有四旋翼的三维（3D）工作空间，提出了一种多飞行高度（MFL）方法来提高探索效率。此外，当四旋翼飞行器到达没有边界的死胡同时，高效回溯算法使用 SSG 提供的图谱选择最佳路径进行有效回溯。凭借这些贡献，我们成功地为多个四旋翼飞行器开发了基于前沿的探索策略，并且通过数值模拟和实验验证了整体方法的性能。