融合语义分割与模糊推理的无人机应急降落选址算法

doi:10.3778/j.issn.1002-8331.2403-0132

摘要/Abstract

摘要： 随着无人机的应用领域从娱乐摄影拓展到物流、军事和灾害响应，对于无人机的自主智能化要求也越来越高。针对无人机紧急情况下自主降落区域复杂难以保证着陆安全的问题，提出了一种实时语义分割网络与模糊推理相结合的降落选址算法（STDC-LSSNet）。考虑到潜在危险因素在航拍图像上占比小、易被错误分割的问题，提出了小目标特征提取模块（small target feature capture module，STFCM），通过计算不同尺度特征的相似性并进行权重分配，强化小目标特征的表达。考虑到安全区域与危险区域边界混淆会导致无人机降落存在巨大风险，提出了边界特征融合模块（boundary feature fusion module，BFFM），将浅层网络由拉普拉斯卷积得到的边界信息与深层网络的语义信息进行特征融合，引入注意力机制，增强边界区域特征的表达。通过对分割得到的图像进行模糊推理，从而精确识别应急降落地点。所提算法在公开数据集Semantic Drone和AeroScapes上与最先进的算法进行了广泛的对比实验，mIoU提升1.72个百分点和3.89个百分点，实时分割速度达到210 FPS，选址的速度达到58.62 ms，实现了无人机在复杂情况下的应急降落选址。

关键词: 无人机, 自主降落, 实时语义分割, 模糊推理, 注意力机制

Abstract: With the expansion of the application fields of drones from recreational photography to logistics, military operations, and disaster response, the demand for the autonomy and intelligence of drones has increased significantly. Addressing the challenge of ensuring autonomous safe landings in complex and unpredictable emergency landing zones, a novel landing site selection algorithm, named STDC-LSSNet (semantic target detection and control for landing site selection net), is proposed by integrating a real-time semantic segmentation network with fuzzy reasoning. Firstly, to address the issue of potential danger factors having a small proportion in aerial images and being prone to missegmentation, a small target feature capture module (STFCM) is introduced. This module calculates the similarity of features at different scales and assigns weights to enhance the representation of small target features effectively. Secondly, considering the risk associated with the confusion between safe and hazardous areas during drone landing due to unclear boundaries, a boundary feature fusion module (BFFM) is introduced. This module combines the boundary information obtained from Laplacian convolution in shallow networks with the semantic information from deep networks, incorporating an attention mechanism to reinforce the representation of boundary region features. Finally, fuzzy reasoning is performed on the segmented images to accurately identify the emergency landing location. The proposed method is extensively evaluated on public datasets, Semantic Drone and AeroScapes, and compared with state-of-the-art algorithms. The results show a significant improvement in mean intersection over union (mIoU) by 1.72 percentage points and 3.89 percentage points. The real-time segmentation speed reaches 210?FPS, and the selection speed achieves 58.62 ms, demonstrating the effectiveness of the proposed approach in enabling drones to autonomously select emergency landing sites in complex scenarios.

Key words: unmanned aerial vehicle (UAV), autonomous landing, real-time semantic segmentation, fuzzy reasoning, attention mechanism

李迪, 肖敏, 任东, 谢咏昶, 姚远. 融合语义分割与模糊推理的无人机应急降落选址算法[J]. 计算机工程与应用, 2025, 61(1): 174-185.

LI Di, XIAO Min, REN Dong, XIE Yongchang, YAO Yuan. Integration of Semantic Segmentation and Fuzzy Reasoning for Unmanned Aerial Vehicle Emergency Landing Site Selection Algorithm[J]. Computer Engineering and Applications, 2025, 61(1): 174-185.

参考文献

[1] ZHU Y, PEI H, WANG L, et al. A vision-based autonomous landing method on mobile platform for UAV[C]//Proceedings of the 2023 42nd Chinese Control Conference, Nanjing, China, Jul 19-21, 2023. New York: IEEE, 2023: 4089-4094.
[2] CHEN C, CHEN S, HU G, et al. An auto-landing strategy based on pan-tilt based visual serving for unmanned aerial vehicle in GNSS-denied environments[J]. Aerospace Science and Technology, 2021, 116: 106891.
[3] 李丹, 邓飞, 赵良玉, 等. 基于深度学习的无人机自主降落标识检测方法[J]. 航空兵器, 2023, 30(5): 115-120.
LI D, DENG F, ZHAO L Y, et al. Detect on method of autonomous landing marker for UAV based on deep learning[J]. Aero Weaponry, 2023, 30(5): 115-120.
[4] CHEN J, ZHANG Y, LI J, et al. Integrated air-ground vehicles for UAV emergency landing based on graph convolution network[J]. IEEE Internet of Things Journal, 2021, 9(12): 9106-9116.
[5] 于雷健. 基于深度学习视觉导引的无人机路径规划及着陆系统研究[D]. 青岛: 中国石油大学, 2020.
YU L J. Deep learning oriented vision guidance systems for unmanned aerial vehicle path planning and landing[D]. Qingdao: China University of Petroleum, 2020.
[6] 张松松. 基于视觉的无人机自主着陆设计[D]. 呼和浩特: 内蒙古工业大学, 2022.
ZHANG S S. Vision-based design for autonomous UAV landing[D]. Hohhot: Inner Mongolia University of Technology, 2022.
[7] RONNEBERGER O, FISCHER P, BROX T. U-net: convolutional networks for biomedical image segmentation[C]//Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention, Munich, Germany, Oct 5-9, 2015. Cham: Springer International Publishing, 2015: 234-241.
[8] PASZKE A, CHAURASIA A, KIM S, et al. ENet: a deep neural network architecture for real-time semantic segmentation[J]. arXiv:1606.02147, 2016.
[9] YU C, WANG J, PENG C, et al. BiSeNet: bilateral segmentation network for real-time semantic segmentation[C]//Proceedings of the European Conference on Computer Vision, Munich, Germany, Sep 8-14, 2018. Cham: Springer, 2018: 325-341.
[10] YU C, GAO C, WANG J, et al. BiSeNet v2: bilateral network with guided aggregation for real-time semantic segmentation[J]. International Journal of Computer Vision, 2021, 129: 3051-3068.
[11] CHEN L C, ZHU Y, PAPANDREOU G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//Proceedings of the European Conference on Computer Vision, Munich, Germany, Sep 8-14, 2018. Cham: Springer, 2018: 801-818.
[12] PENG J, LIU Y, TANG S, et al. PP-LiteSeg: a superior real-time semantic segmentation model[J]. arXiv:2204.02681, 2022.
[13] XU J, XIONG Z, BHATTACHARYYA S P. PIDNet: a real-time semantic segmentation network inspired by PID controllers[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Rio de Janeiro, Brazil, Jun 19-24, 2023. New York: IEEE, 2023: 19529-19539.
[14] 尚克军, 郑辛, 王旒军, 等. 基于图像语义分割的无人机自主着陆导航方法[J]. 中国惯性技术学报, 2020, 28(5): 586-594.
SHANG K J, ZHENG X, WANG L J, et al. Image semantic segmentation-based navigation method for UAV auto-landing[J]. Journal of Chinese Inertial Technology, 2020, 28(5): 586-594.
[15] PUTRANTO H Y, IRFANSYAH A N, ATTAMIMI M. Identification of safe landing areas with semantic segmentation and contour detection for delivery UAV[C]//Proceedings of the 2022 9th International Conference on Information Technology, Computer, and Electrical Engineering, Bali, Indonesia, Oct 5-6, 2022. New York: IEEE, 2022: 254-257.
[16] LIU S, QI L, QIN H, et al. Path aggregation network for instance segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, Jun 18-22, 2018. New York: IEEE, 2018: 8759-8768.
[17] KIRILLOV A, GIRSHICK R, HE K, et al. Panoptic feature pyramid networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, Jun 16-20, 2019. New York: IEEE, 2019: 6399-6408.
[18] CHENG B, COLLINS M D, ZHU Y, et al. Panoptic-DeepLab: a simple, strong, and fast baseline for bottom-up panoptic segmentation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, USA, Jun 13-19, 2020. New York: IEEE, 2020: 12475-12485.
[19] PANDA M. Road boundary detection using 3D-to-2D transformation of LIDAR data and conditional generative adversarial networks[C]//Proceedings of the 2020 11th International Conference on Computing, Communication and Networking Technologies, Kharagpur, India, Jul 1-3, 2020. New York: IEEE, 2020: 1-6.
[20] WU J, XU H, SUN R, et al. Road boundary-enhanced automatic background filtering for roadside LiDAR sensors[J]. IEEE Intelligent Transportation Systems Magazine, 2021, 14(4): 60-72.
[21] UDDIN N, HERMAWAN H, SITORUS F J P, et al. Lane detection system based on canny method for driving assistance[C]//Proceedings of the 2023 International Conference on Advanced Mechatronics, Intelligent Manufacture and Industrial Automation, Surabaya, Indonesia, Oct 17-18, 2023. New York: IEEE, 2023: 1-6.
[22] 冯樱, 乔宝山, 江子旺, 等. 基于模糊推理的换道决策与仿真验证[J]. 重庆交通大学学报 (自然科学版), 2023, 42(6): 155-162.
FENG Y, QIAO B S, JIANG Z W, et al. Lane change decision and simulation verification based on fuzzy reasoning[J]. Journal of Chongqing Jiaotong University (Natural Science), 2023, 42(6): 155-162.
[23] 林鸿裕, 卢桂宁. 基于模糊逻辑综合评判的垃圾处理厂选址仿真[J]. 计算机仿真, 2024, 41(1): 513-517.
LIN H Y, LU G N. Simulation of waste treatment plant location based on fuzzy logic synthesis comprehensive evaluation[J]. Computer Simulation, 2024, 41(1): 513-517.
[24] 刘建娟, 刘忠璞, 张会娟, 等. 基于模糊控制蚁群算法的移动机器人路径规划[J]. 组合机床与自动化加工技术, 2023(1): 20-24.
LIU J J, LIU Z P, ZHANG H J, et al. Path planning of mobile robot based on fuzzy control ant colony algorithm[J]. Modular Machine Tool & Automatic Manufacturing Technique, 2023(1): 20-24.
[25] FAN M, LAI S, HUANG J, et al. Rethinking BiSeNet for real-time semantic segmentation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, Jun 19-25, 2021. New York: IEEE, 2021: 9716-9725.
[26] NIGAM I, HUANG C, RAMANAN D. Ensemble knowledge transfer for semantic segmentation[C]//Proceedings of the 2018 IEEE Winter Conference on Applications of Computer Vision, Lake Tahoe, USA, Mar 12-15, 2018. New York: IEEE, 2018: 1499-1508.
[27] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision, Munich, Germany, Sep 8-14, 2018. Cham: Springer, 2018: 3-19.