Hongyu Song,Jincheng Yu,Jiantao Qiu,Zhixiao Sun,Kuijun Lang,Qing Luo,Yuan Shen,Yu Wang
Abstract:For scenes such as floods and earthquakes, the disaster area is large, and rescue time is tight. Multi-UAV exploration is more efficient than a single UAV. Existing UAV exploration work is modeled as a Coverage Path Planning (CPP) task to achieve full coverage of the area in the presence of obstacles. However, the endurance capability of UAV is limited, and the rescue time is urgent. Thus, even using multiple UAVs cannot achieve complete disaster area coverage in time. Therefore, in this paper we propose a multi-Agent Endurance-limited CPP (MAEl-CPP) problem based on a priori heatmap of the disaster area, which requires the exploration of more valuable areas under limited energy. Furthermore, we propose a path planning algorithm for the MAEl-CPP problem, by ranking the possible disaster areas according to their importance through satellite or remote aerial images and completing path planning according to the importance level. Experimental results show that our proposed algorithm is at least twice as effective as the existing method in terms of search efficiency.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is how to preferentially cover the most valuable areas under the limited endurance of unmanned aerial vehicles (UAVs) when conducting efficient exploration in disaster - stricken areas. Specifically, the paper defines a multi - agent endurance - limited coverage path planning (MAEl - CPP) problem to guide the exploration tasks of UAVs based on the prior heat map of the disaster area. This problem is proposed because although the existing multi - UAV exploration methods can improve the search efficiency, in practical applications, due to the limitations of UAV endurance and rescue time, it is difficult to achieve complete coverage of the entire disaster area. Therefore, by preferentially exploring high - value areas, the chances of trapped people or affected properties being rescued and protected in a timely manner can be increased.
### Background and Problem Description of the Paper
In disaster - stricken areas such as floods and earthquakes, rescue time is very urgent, and these areas are usually large in area, making it difficult for a single UAV to complete a comprehensive exploration task. Therefore, multi - UAV exploration has become a more effective solution. However, multi - UAV exploration faces several challenges:
1. **Limited Endurance of UAVs**: The flight time of each UAV is limited, which restricts the area range that it can cover.
2. **Urgent Rescue Time**: After a disaster occurs, the rescue time is very limited, and it is necessary to complete the exploration of key areas in a short time.
3. **Limitations of Existing Methods**: Traditional coverage path planning (CPP) methods mainly focus on complete map coverage and do not take into account the limitations of UAV endurance and rescue time.
### Solutions Proposed in the Paper
To address the above challenges, the paper proposes the following solutions:
1. **Define the MAEl - CPP Problem**: Based on the prior heat map of the disaster area, define a multi - agent endurance - limited coverage path planning problem. The heat map reflects the value of different areas and helps UAVs preferentially explore high - value areas.
2. **Generate Heat Map**: Use satellite images or remote aerial photography images to generate heat maps, and convert different scenes (such as houses, trees, grasslands, water bodies, etc.) into corresponding heat values through classification algorithms.
3. **Path Planning Algorithms**: Propose several path planning algorithms suitable for MAEl - CPP, among which the SVReC algorithm stands out in terms of search efficiency, reaching 4.2 times that of the traditional Zig - Zag search algorithm.
### Experimental Results
The paper verifies the effectiveness of the proposed method through experiments. The experimental results show that the SVReC algorithm exhibits higher search efficiency and coverage in both single - UAV and multi - UAV cases. Specifically:
- **Single - UAV Experiment**: On a small - scale map, the cumulative Eff - weight of the SVReC algorithm is significantly higher than that of other algorithms. Although its coverage may be slightly lower, its search efficiency is higher.
- **Multi - UAV Experiment**: As the map scale increases, the advantages of the SVReC algorithm become more obvious. Its cumulative Eff - weight is much higher than that of other algorithms, and at the same time, the coverage rate is also relatively high.
- **Search Time Evaluation**: The SVReC algorithm requires significantly less time to accumulate the same Eff - weight than the Zig - Zag algorithm, especially in the multi - UAV case.
### Conclusion
By defining the MAEl - CPP problem and proposing corresponding path planning algorithms, the paper effectively solves the challenge of preferentially covering high - value areas under the limited endurance of UAVs. The experimental results show that the SVReC algorithm performs excellently in terms of search efficiency and coverage and has high practical value. Future work can further optimize the algorithm to improve its adaptability and robustness in complex environments.