Fast and Communication-Efficient Multi-UAV Exploration Via Voronoi Partition on Dynamic Topological Graph

TL;DR

The paper tackles the challenge of scalable, communication-efficient multi-UAV exploration in large, cluttered environments. It introduces a fast-sharing multi-robot dynamic topological graph (MR-DTG) and utilizes graph Voronoi partitioning on MR-DTG to achieve decentralized, cost-aware task allocation, coupled with a two-level exploration planner and MINCO trajectories. The approach significantly reduces data transmission and speeds up exploration, as demonstrated by simulations showing up to ~38% faster exploration and ~95% less communication, and by real-world tests with 6 UAVs. These results suggest a practical, scalable solution for collaborative aerial exploration and mapping, with plans to release open-source code.

Abstract

Efficient data transmission and reasonable task allocation are important to improve multi-robot exploration efficiency. However, most communication data types typically contain redundant information and thus require massive communication volume. Moreover, exploration-oriented task allocation is far from trivial and becomes even more challenging for resource-limited unmanned aerial vehicles (UAVs). In this paper, we propose a fast and communication-efficient multi-UAV exploration method for exploring large environments. We first design a multi-robot dynamic topological graph (MR-DTG) consisting of nodes representing the explored and exploring regions and edges connecting nodes. Supported by MR-DTG, our method achieves efficient communication by only transferring the necessary information required by exploration planning. To further improve the exploration efficiency, a hierarchical multi-UAV exploration method is devised using MR-DTG. Specifically, the \emph{graph Voronoi partition} is used to allocate MR-DTG's nodes to the closest UAVs, considering the actual motion cost, thus achieving reasonable task allocation. To our knowledge, this is the first work to address multi-UAV exploration using \emph{graph Voronoi partition}. The proposed method is compared with a state-of-the-art method in simulations. The results show that the proposed method is able to reduce the exploration time and communication volume by up to 38.3\% and 95.5\%, respectively. Finally, the effectiveness of our method is validated in the real-world experiment with 6 UAVs. We will release the source code to benefit the community.

Figures (8)

• Figure 1: An exploration instance of the proposed multi-UAV exploration, where 15 UAVs are exploring the environment with MR-DTG. (a) The simulation environment of a city scene. (b) Exploration results using the proposed method. Dots represent history nodes, and cubes represent explorable regions. The colors of history nodes correspond to the UAVs they are assigned to during the global partition, and the colors of explorable regions correspond to the UAVs they are assigned to during the local partition. Video of the experiments is available at: https://www.youtube.com/watch?v=AtG9stNVjX0&t=1s.
• Figure 2: Overview of the proposed multi-UAV exploration system.
• Figure 3: The generation of the history node in MR-DTG. (a) Each history node and UAV maintains a Dijkstra tree to calculate the shortest path from the surrounding voxel to it. (b) The current history node will expand its Dijkstra tree to cover the space that the UAV's Dijkstra search covers (green background). (c) A new history node will be generated when the UAV is far away from history nodes. The new history node will be connected to nodes that are covered by its Dijkstra tree. (d) History nodes are connected through handshakes (the orange and purple nodes "handshake" in voxels surrounded by dotted lines). Any one of the shortest paths corresponding to the handshake voxels (voxels surrounded by solid lines) will be stored in the edge between these history nodes.
• Figure 4: A microelement of frontier area $ds$ that the collision-free ray reaches at a distance $r_{i,j}$ in the direction $(\theta_i, \phi_j)$. The angular space within the viewpoint's FoV is evenly divided in the yaw and pitch directions with sizes $\Delta\theta$ and $\Delta\phi$ respectively.
• Figure 5: Comparison of Voronoi partition of Euclidean space and graph Voronoi partition on MR-DTG. (a) The Voronoi partition of Euclidean space assigns the upper target to the yellow UAV and the lower target to the red UAV. Due to the existence of obstacles, both UAVs must take a long detour to reach the target. (b) Graph Voronoi partition on MR-DTG takes into account the traversable path length between nodes, assigning EROI to UAVs with the lowest exploration motion cost.