EDEN: Efficient Dual-Layer Exploration Planning for Fast UAV Autonomous Exploration in Large 3-D Environments

TL;DR

EDEN tackles the challenge of real-time, large-scale UAV exploration by introducing a dual-layer planner that first computes a long-horizon region routing with the exploration-oriented heuristic double-tree (EOHDT) and then selects curvature-aware short-horizon viewpoints to drive aggressive yet safe exploration. The second layer uses an aggressive and safe exploration-oriented (ASEO) trajectory, based on a 4-D MINCO formulation that links exploring, continuous, and safety segments while optimizing feasibility and time. Key contributions include the EOHDT-based global routing with a provable length bound, the curvature-penalized viewpoint scoring for high-speed exploration, and the ASEO trajectory optimization that ensures continuity and safety. Empirical results in diverse simulations and real-world tests show substantial gains in exploration efficiency, UAV speed, and computational efficiency, demonstrating real-time planning capability in hundred-meter-scale environments and practical applicability for large-scale UAV exploration.

Abstract

Efficient autonomous exploration in large-scale environments remains challenging due to the high planning computational cost and low-speed maneuvers. In this paper, we propose a fast and computationally efficient dual-layer exploration planning method. The insight of our dual-layer method is efficiently finding an acceptable long-term region routing and greedily exploring the target in the region of the first routing area with high speed. Specifically, the proposed method finds the long-term area routing through an approximate algorithm to ensure real-time planning in large-scale environments. Then, the viewpoint in the first routing region with the lowest curvature-penalized cost, which can effectively reduce decelerations caused by sharp turn motions, will be chosen as the next exploration target. To further speed up the exploration, we adopt an aggressive and safe exploration-oriented trajectory to enhance exploration continuity. The proposed method is compared to state-of-the-art methods in challenging simulation environments. The results show that the proposed method outperforms other methods in terms of exploration efficiency, computational cost, and trajectory speed. We also conduct real-world experiments to validate the effectiveness of the proposed method. The code will be open-sourced.

Figures (6)

• Figure 1: The overview of the proposed method framework.
• Figure 2: An illustration of DTG and an instance of EOHDT-based routing planning. (a) DTG and its boundary regions (pink dashed boxes). (b) An MST that connects all the boundary regions of DTG is extracted. (c) The leaf point, whose distance is the longest to the root node, and its corresponding branch (green) will be executed last. The numbers indicate the lengths of edges. (d) Once there are multiple child nodes for the current node, the child node (orange dashed curve) whose first node has the shortest distance to the current node will be chosen as the next branch. (e) The final routing found by the proposed EOHDT-based routing planning.
• Figure 3: An instance of ASEO trajectory. The initial position, velocity, and acceleration of $\mathcal{J}_e$ (violet) are fixed with the current UAV position, velocity, and acceleration (yellow). The continuous trajectory $\mathcal{J}_c$ (red) ends at $\mathbf{vp}_c'$. The safety trajectory $\mathcal{J}_s$ (blue) ends at $\mathbf{vp}_s'$. The optimizable parameters, including intermediate 4-D points, trajectory durations, velocity, and acceleration at $\mathbf{vp}_e'$, are illustrated in green.
• Figure 4: The mapping results and the executed trajectory of all the methods. The color of the trajectory (j) denotes the velocity of the UAV. Each row illustrates the simulations in one environment. The last row illustrates the explored volume vs. time.
• Figure 5: Simulation result in $\textit{Large Tunnel}$.

Theorems & Definitions (4)

• Theorem 1
• proof
• Theorem 2
• proof