Ocean front, internal wave and ocean vortex are general marine physical phenomena. The traditional observation method is from the angle of image gray level, even the existing deep learning network is based on image gray level through line detection. In this paper, a new observation method is proposed, that is, the retrieved Doppler anomaly of ocean wave motion relative to satellite antenna. For estimating Doppler anomaly, a new algorithm is proposed, which is based on Bayesian estimation method and reaches Cramer boundary through iteration. To verify the effectiveness of the algorithm, this paper uses GaoFen-3 SLC (single look complex image) SAR image. The results of local radial velocity distribution of inversion results are analyzed. The gradient distribution of local radial velocity, that is, the place where the velocity of ocean front changes the most, has the largest change in velocity gradient, which can better explain the wave modulation effect in ocean physics. Compared with the conventional method, our method can better understand and explain the marine physical phenomena by retrieving the radial velocity of ocean current.
{"title":"Observing Ocean Front by Retrieving Doppler Anomaly from GaoFen-3 SAR Images","authors":"J. Wang, Yanlang Xu, Xiaoqing Wang, Boting Pan, Mingkai Tao, Haifeng Huang","doi":"10.1109/ICUS55513.2022.9987205","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9987205","url":null,"abstract":"Ocean front, internal wave and ocean vortex are general marine physical phenomena. The traditional observation method is from the angle of image gray level, even the existing deep learning network is based on image gray level through line detection. In this paper, a new observation method is proposed, that is, the retrieved Doppler anomaly of ocean wave motion relative to satellite antenna. For estimating Doppler anomaly, a new algorithm is proposed, which is based on Bayesian estimation method and reaches Cramer boundary through iteration. To verify the effectiveness of the algorithm, this paper uses GaoFen-3 SLC (single look complex image) SAR image. The results of local radial velocity distribution of inversion results are analyzed. The gradient distribution of local radial velocity, that is, the place where the velocity of ocean front changes the most, has the largest change in velocity gradient, which can better explain the wave modulation effect in ocean physics. Compared with the conventional method, our method can better understand and explain the marine physical phenomena by retrieving the radial velocity of ocean current.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131849424","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
As an important support for modern air combat intelligent auxiliary decision-making, real-time and high-precision target intent recognition addresses the foundation for realizing deep situational awareness and creating tactical opportunities. Aiming at the limitation of the existing algorithms such as dependence on empirical knowledge, difficulty in extracting the full temporal characteristics, and inability to meet the requirements of actual air combat, this paper proposes a target tactical intention recognition algorithm based on bi-directional Long Short-Term Memory (BiLSTM). Firstly, we analyze the air combat mechanism to construct the target tactical intention space based on the tactical layer. Specifically, suitable characteristics are selected to describe the intention space. We then design a recognition method considering the characteristic of the tactical intention space. Finally, compared with other algorithms, the simulation results show the effectiveness of the proposed method, which outperforms other methods in terms of accuracy at 92%. And the results are more practical.
{"title":"Tactical Intention Recognition Method of Air Combat Target Based on BiLSTM network","authors":"Xingyu Wang, Zhen Yang, Guang Zhan, Jichuan Huang, Shiyuan Chai, Deyun Zhou","doi":"10.1109/ICUS55513.2022.9986667","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9986667","url":null,"abstract":"As an important support for modern air combat intelligent auxiliary decision-making, real-time and high-precision target intent recognition addresses the foundation for realizing deep situational awareness and creating tactical opportunities. Aiming at the limitation of the existing algorithms such as dependence on empirical knowledge, difficulty in extracting the full temporal characteristics, and inability to meet the requirements of actual air combat, this paper proposes a target tactical intention recognition algorithm based on bi-directional Long Short-Term Memory (BiLSTM). Firstly, we analyze the air combat mechanism to construct the target tactical intention space based on the tactical layer. Specifically, suitable characteristics are selected to describe the intention space. We then design a recognition method considering the characteristic of the tactical intention space. Finally, compared with other algorithms, the simulation results show the effectiveness of the proposed method, which outperforms other methods in terms of accuracy at 92%. And the results are more practical.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"106 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132196703","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Convolutional neural network has excellent representation learning ability, which makes it unique in the field of behavior prediction. This paper presents a prediction method of pedestrian behavior around intelligent vehicles, which makes use of the advantages of convolutional neural network, combines pedestrian intention and road environment information to predict pedestrian behavior around intelligent vehicle, and optimizes pedestrian avoidance module in automatic driving system. The experimental results show that the area of the closed graph composed of the predicted trajectory and the actual trajectory is 0.1269 $m$2. The method proposed in this paper can effectively predict the pedestrian behavior trajectory, ensure the safety of drivers and pedestrians to the maximum extent, and provide a new solution for intelligent vehicles and intelligent driving path planning.
{"title":"Pedestrian Behavior Prediction Method for Intelligent Vehicles Based on Convolutional Neural Network","authors":"Hongbo Gao, Xi He, Liuchang Wang, Fei Zhang, Kaiquan Cai, Xiaozhao Fang","doi":"10.1109/ICUS55513.2022.9987009","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9987009","url":null,"abstract":"Convolutional neural network has excellent representation learning ability, which makes it unique in the field of behavior prediction. This paper presents a prediction method of pedestrian behavior around intelligent vehicles, which makes use of the advantages of convolutional neural network, combines pedestrian intention and road environment information to predict pedestrian behavior around intelligent vehicle, and optimizes pedestrian avoidance module in automatic driving system. The experimental results show that the area of the closed graph composed of the predicted trajectory and the actual trajectory is 0.1269 $m$2. The method proposed in this paper can effectively predict the pedestrian behavior trajectory, ensure the safety of drivers and pedestrians to the maximum extent, and provide a new solution for intelligent vehicles and intelligent driving path planning.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131867998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-28DOI: 10.1109/ICUS55513.2022.9987243
Zhiwen Huang, Hongxu Li, Yuanfu Zhong, Qian Su, Na Lv, Yingchao Zhang
Historical war experience demonstrates that crashing critical supply facility has become an efficient tactic during wartime, so it is of great strategic significance to design an efficient and reliable emergency logistics network in face of the uncertainty at war. This paper presents a scenario-based modelling method by considering both demand uncertainty and facility failure scenarios in the mathematical model and the immune genetic algorithm (IGA) is applied to solve it. A set of numerical experiments has been conducted to verify the model and effectiveness of IGA. Finally, the sensitivity analysis results show facility failure probability has a greater impact on the final location decision, which provide model and approach for the location-allocation problem during wartime.
{"title":"Reliability Facility Location with Fuzzy Demand and Failure Scenarios","authors":"Zhiwen Huang, Hongxu Li, Yuanfu Zhong, Qian Su, Na Lv, Yingchao Zhang","doi":"10.1109/ICUS55513.2022.9987243","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9987243","url":null,"abstract":"Historical war experience demonstrates that crashing critical supply facility has become an efficient tactic during wartime, so it is of great strategic significance to design an efficient and reliable emergency logistics network in face of the uncertainty at war. This paper presents a scenario-based modelling method by considering both demand uncertainty and facility failure scenarios in the mathematical model and the immune genetic algorithm (IGA) is applied to solve it. A set of numerical experiments has been conducted to verify the model and effectiveness of IGA. Finally, the sensitivity analysis results show facility failure probability has a greater impact on the final location decision, which provide model and approach for the location-allocation problem during wartime.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133822615","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Unmanned aerial vehicles (UAVs) path planning refers to exploring the optimal flight trajectory from the starting point to the destination that satisfies the UAV under specific constraints such as maneuverability and environmental information constraints, which is a crucial technology for the UAV mission planning. In order to enhance the efficiency and safety of the UAV path planning task, a new autonomous UAV path planning system based on deep reinforcement learning is proposed in this article. At the beginning, a new action guidance strategy based on the Deep Q-Network (DQN) algorithm is introduced via deploying the Gravity-aware Deep Q-Network (Ga-DQN) method. This strategy can effectively assist the UAVs to avoid the obstacles in the specific state. For balancing the efficiency and safety of the task, a reward scheme that introduces a safety counting mechanism is proposed to provide global guidance for the agent in Deep Reinforcement Learning (DRL). The simulation results under different obstacle densities show that the proposed novel strategy can obviously behave robust and greater efficiency compared to the traditional methods.
{"title":"Ga-DQN: A Gravity-aware DQN Based UAV Path Planning Algorithm","authors":"Zhicheng Xu, Qi Wang, Fuchen Kong, Hualong Yu, Shang Gao, Demin Pan","doi":"10.1109/ICUS55513.2022.9986557","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9986557","url":null,"abstract":"Unmanned aerial vehicles (UAVs) path planning refers to exploring the optimal flight trajectory from the starting point to the destination that satisfies the UAV under specific constraints such as maneuverability and environmental information constraints, which is a crucial technology for the UAV mission planning. In order to enhance the efficiency and safety of the UAV path planning task, a new autonomous UAV path planning system based on deep reinforcement learning is proposed in this article. At the beginning, a new action guidance strategy based on the Deep Q-Network (DQN) algorithm is introduced via deploying the Gravity-aware Deep Q-Network (Ga-DQN) method. This strategy can effectively assist the UAVs to avoid the obstacles in the specific state. For balancing the efficiency and safety of the task, a reward scheme that introduces a safety counting mechanism is proposed to provide global guidance for the agent in Deep Reinforcement Learning (DRL). The simulation results under different obstacle densities show that the proposed novel strategy can obviously behave robust and greater efficiency compared to the traditional methods.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133387867","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Based on the decision-making architecture of information pooling and sharing in the hidden layer, the communication protocol is set manually, and the pooling method is used to integrate the information. Although the problem of communication and extension between agents is solved, it is difficult for tasks lacking prior knowledge to design effective communication protocols. The centralized decision- making architecture based on two-way RNN communication uses the information storage characteristics of two-way RNN structure. It can self learn the communication protocol between agents, which overcomes the rigid requirement of task prior knowledge in communication protocol design. The action distribution of a single agent is used as the output of the multi- agent network to replace the joint action distribution, and the global state information in the environment is used as the input instead of simply inputting the local information to different agents. The effectiveness of the method is verified by an example.
{"title":"Decision-making Method Based on Multi-agent Deep Reinforcement Learning","authors":"Weiwei Bian, Chunguang Wang, Chan Liu, Kuihua Huang, Ying Mi, Yanxiang Jia","doi":"10.1109/ICUS55513.2022.9987201","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9987201","url":null,"abstract":"Based on the decision-making architecture of information pooling and sharing in the hidden layer, the communication protocol is set manually, and the pooling method is used to integrate the information. Although the problem of communication and extension between agents is solved, it is difficult for tasks lacking prior knowledge to design effective communication protocols. The centralized decision- making architecture based on two-way RNN communication uses the information storage characteristics of two-way RNN structure. It can self learn the communication protocol between agents, which overcomes the rigid requirement of task prior knowledge in communication protocol design. The action distribution of a single agent is used as the output of the multi- agent network to replace the joint action distribution, and the global state information in the environment is used as the input instead of simply inputting the local information to different agents. The effectiveness of the method is verified by an example.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132812593","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-28DOI: 10.1109/ICUS55513.2022.9986900
Chengcheng Wang, Yulong Wang, Chen Peng
The purpose of this paper is to discuss distributed cooperative target tracking for a multi-unmanned surface vehicle (multi-USV) system. The cooperative target tracking problem is formulated as a multi-USV learning problem. Based on this formulation, a multi-USV distributed cooperative target tracking (MUTT) algorithm is proposed. To avoid the collisions between USVs during the tracking process, an additional safety layer is introduced. Some safety signals are constructed based on USVs' states. By correcting actions through the trained safety layer, USVs can avoid collisions reasonably. Moreover, for the sake of demonstrating the effectiveness of the proposed MUTT algorithm in target tracking, reward functions and mission scenarios are well constructed. Furthermore, a comparison of the MUTT algorithm and Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm is given. The obtained results manifest that the proposed MUTT algorithm provides safe policies for multi-USV cooperative target tracking tasks.
{"title":"Multi-USV Deep Reinforcement Learning for Distributed Cooperative Target Tracking","authors":"Chengcheng Wang, Yulong Wang, Chen Peng","doi":"10.1109/ICUS55513.2022.9986900","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9986900","url":null,"abstract":"The purpose of this paper is to discuss distributed cooperative target tracking for a multi-unmanned surface vehicle (multi-USV) system. The cooperative target tracking problem is formulated as a multi-USV learning problem. Based on this formulation, a multi-USV distributed cooperative target tracking (MUTT) algorithm is proposed. To avoid the collisions between USVs during the tracking process, an additional safety layer is introduced. Some safety signals are constructed based on USVs' states. By correcting actions through the trained safety layer, USVs can avoid collisions reasonably. Moreover, for the sake of demonstrating the effectiveness of the proposed MUTT algorithm in target tracking, reward functions and mission scenarios are well constructed. Furthermore, a comparison of the MUTT algorithm and Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm is given. The obtained results manifest that the proposed MUTT algorithm provides safe policies for multi-USV cooperative target tracking tasks.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115030681","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper compares several Ambiguity Resolution (AR) methods, including the Least square AMBiguity Decorrelation Adjustment (LAMBDA) method, the Modified LAMBDA (MLAMBDA) method, the Two-step Success Rate Criterion (TSRC) method with the low cost GNSS receivers. The algorithms were firstly tested with the low cost ublox F9P dual frequency multi-GNSS receivers in vehicle field test and UAV flight test. The ambiguity fix rate and the Time To First Fix (TTFF) are used as indices to compare the algorithms. Experiments show that the Co-LAMBDA algorithm achieves a TTFF of 480 epochs and a fix rate of 53.7%, and the TSRC algorithm achieves a TTFF of 112 epochs and a fix rate of 91.3%. It can be seen that TSRC algorithm has better performance in both TTFF and fix rate in the low cost GNSS UAV dynamic positioning applications. Then the algorithms were tested with a quasi-dynamic medium-long baseline Real-Time Kinematic (RTK) experiment, a total of 970 experimental results verify that the TSRC algorithm improves the median fix rate from 41.51% to 90.83%, and the median correct rate slightly degrades from 100% to 96.98%, which is reasonable since it computes the statistics from many more fixed-solution samples.
本文将最小二乘歧义去相关平差法(LAMBDA)、改进LAMBDA法(MLAMBDA)、两步成功率准则法(TSRC)等几种歧义解决方法与低成本GNSS接收机进行了比较。首先在低成本ublox F9P双频多gnss接收机上进行了车辆现场试验和无人机飞行试验。以歧义修复率和首次修复时间(Time To First fix, TTFF)作为比较算法的指标。实验表明，Co-LAMBDA算法的TTFF为480个epoch，固定率为53.7%，TSRC算法的TTFF为112个epoch，固定率为91.3%。可以看出，在低成本GNSS无人机动态定位应用中，TSRC算法在TTFF和固定率方面都具有更好的性能。然后通过准动态中长期基线实时运动学(RTK)实验对算法进行了测试，共970个实验结果验证，TSRC算法将中位数固定率从41.51%提高到90.83%，中位数正确率从100%略微下降到96.98%，这是合理的，因为它计算了更多固定解样本的统计量。
{"title":"Partial Ambiguity Resolution for Low Cost GNSS Receiver in UAV Navigation Applications: A Comparative Study","authors":"Xin Liu, Jiaju Guo, Haoli Zhang, Dezhong Zhou, Yanqing Hou","doi":"10.1109/ICUS55513.2022.9987056","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9987056","url":null,"abstract":"This paper compares several Ambiguity Resolution (AR) methods, including the Least square AMBiguity Decorrelation Adjustment (LAMBDA) method, the Modified LAMBDA (MLAMBDA) method, the Two-step Success Rate Criterion (TSRC) method with the low cost GNSS receivers. The algorithms were firstly tested with the low cost ublox F9P dual frequency multi-GNSS receivers in vehicle field test and UAV flight test. The ambiguity fix rate and the Time To First Fix (TTFF) are used as indices to compare the algorithms. Experiments show that the Co-LAMBDA algorithm achieves a TTFF of 480 epochs and a fix rate of 53.7%, and the TSRC algorithm achieves a TTFF of 112 epochs and a fix rate of 91.3%. It can be seen that TSRC algorithm has better performance in both TTFF and fix rate in the low cost GNSS UAV dynamic positioning applications. Then the algorithms were tested with a quasi-dynamic medium-long baseline Real-Time Kinematic (RTK) experiment, a total of 970 experimental results verify that the TSRC algorithm improves the median fix rate from 41.51% to 90.83%, and the median correct rate slightly degrades from 100% to 96.98%, which is reasonable since it computes the statistics from many more fixed-solution samples.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115047974","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-28DOI: 10.1109/ICUS55513.2022.9986971
Ya Wang, Lei Shi, Jinliang Shao, Yuhua Cheng, Houjun Wang
This paper investigates the issue of localization for wireless sensor networks resistant to denial-of-service (DoS) attacks with the assumption that each attack consists of an active period and a dormant period due to limited power. On the basis of barycentric coordinates involving relative distance measurements, a hold-on strategy based distributed localization (HS-DILOC) algorithm is proposed. Explicitly when the communication channel of a sensor is perpetrated by DoS attacks, HS-DILOC allows the sensor to update its coordinates utilizing the previous packets collected from its neighbors during the last dormant period. In addition, this paper theoretically shows that the proposed algorithm is capable of converging to the accurate locations of sensors disregarding the attack strategy. Finally, the experiments on Raspberry Pis are used to illustrate the validity of the proposed algorithm.
{"title":"Distributed Iterative Localization for Wireless Sensor Networks Subject to DoS Attacks","authors":"Ya Wang, Lei Shi, Jinliang Shao, Yuhua Cheng, Houjun Wang","doi":"10.1109/ICUS55513.2022.9986971","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9986971","url":null,"abstract":"This paper investigates the issue of localization for wireless sensor networks resistant to denial-of-service (DoS) attacks with the assumption that each attack consists of an active period and a dormant period due to limited power. On the basis of barycentric coordinates involving relative distance measurements, a hold-on strategy based distributed localization (HS-DILOC) algorithm is proposed. Explicitly when the communication channel of a sensor is perpetrated by DoS attacks, HS-DILOC allows the sensor to update its coordinates utilizing the previous packets collected from its neighbors during the last dormant period. In addition, this paper theoretically shows that the proposed algorithm is capable of converging to the accurate locations of sensors disregarding the attack strategy. Finally, the experiments on Raspberry Pis are used to illustrate the validity of the proposed algorithm.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"92 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123417089","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-10-28DOI: 10.1109/ICUS55513.2022.9986765
Yan Zhang, Yucan Chi, Yongsheng Fan
With the rapid development of space remote sensing technology, accurate ship detection based on high-resolution optical remote sensing images has steadily attracted considerable research interest. However, most of the current methods adopt a fixed horizontal detection frame to predict the target. Although these methods have good detection accuracy, because the ship's orientation is arbitrary in reality, a large error occurs in the matching degree of the detection effective area, resulting in inaccurate target detection. Therefore, this paper proposes a ship detection algorithm based on an arbitrary quadrilateral prediction frame. We redefine the loss function and directly predict the detection frame's four vertices through the designed eight-parameter regression process. In addition, the convolutional block attention module (CBAM) is introduced to optimize the original network structure, and the clustering method is used to optimize the calculation of the anchor point. To replace the intersection over union (IoU), which cannot distinguish different alignments of objects, we adopt a generalized intersection over union (GIoU). Finally, we conduct experiments based on the DOTA ship dataset and the HRSC2016 dataset. The results show that our method is better than YOLOv3 and other commonly used target detection algorithms in terms of accuracy and visualization. Meanwhile, we compared with SOTA algorithm in real-time and dense ship detection. Experimental results prove that its speed and performance on mobile platform are in the lead, and it has a great effect on dense ship detection.
随着空间遥感技术的快速发展，基于高分辨率光学遥感图像的船舶精确探测日益引起人们的关注。然而，目前的方法大多采用固定的水平检测帧来预测目标。这些方法虽然具有较好的检测精度，但由于现实中舰船的方位是任意的，在检测有效区域的匹配程度上出现较大误差，导致目标检测不准确。为此，本文提出了一种基于任意四边形预测框架的船舶检测算法。我们重新定义损失函数，通过设计的八参数回归过程直接预测检测帧的四个顶点。此外，引入卷积块注意力模块(CBAM)对原有网络结构进行优化，并采用聚类方法对锚点的计算进行优化。为了取代不能区分物体不同排列的交并(intersection over union, IoU)，我们采用广义交并(GIoU)。最后，我们基于DOTA船舶数据集和HRSC2016数据集进行了实验。结果表明，我们的方法在精度和可视化方面都优于YOLOv3和其他常用的目标检测算法。同时，比较了SOTA算法在实时和密集船舶检测方面的性能。实验结果表明，该方法在移动平台上的速度和性能都处于领先地位，对密集船舶检测有很大的效果。
{"title":"Highly Adaptive Ship Detection Based on Arbitrary Quadrilateral Bounding Box","authors":"Yan Zhang, Yucan Chi, Yongsheng Fan","doi":"10.1109/ICUS55513.2022.9986765","DOIUrl":"https://doi.org/10.1109/ICUS55513.2022.9986765","url":null,"abstract":"With the rapid development of space remote sensing technology, accurate ship detection based on high-resolution optical remote sensing images has steadily attracted considerable research interest. However, most of the current methods adopt a fixed horizontal detection frame to predict the target. Although these methods have good detection accuracy, because the ship's orientation is arbitrary in reality, a large error occurs in the matching degree of the detection effective area, resulting in inaccurate target detection. Therefore, this paper proposes a ship detection algorithm based on an arbitrary quadrilateral prediction frame. We redefine the loss function and directly predict the detection frame's four vertices through the designed eight-parameter regression process. In addition, the convolutional block attention module (CBAM) is introduced to optimize the original network structure, and the clustering method is used to optimize the calculation of the anchor point. To replace the intersection over union (IoU), which cannot distinguish different alignments of objects, we adopt a generalized intersection over union (GIoU). Finally, we conduct experiments based on the DOTA ship dataset and the HRSC2016 dataset. The results show that our method is better than YOLOv3 and other commonly used target detection algorithms in terms of accuracy and visualization. Meanwhile, we compared with SOTA algorithm in real-time and dense ship detection. Experimental results prove that its speed and performance on mobile platform are in the lead, and it has a great effect on dense ship detection.","PeriodicalId":345773,"journal":{"name":"2022 IEEE International Conference on Unmanned Systems (ICUS)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121384212","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：