Traditional meta reinforcement learning based on task inference separates task inference with task control, but ignores the importance of exploration during task inference. The agent uses the same policy for both task exploration process and task control process, which leads to low task inference efficiency. To solve this problem, this paper proposes a task inference based meta reinforcement learning framework (Separating Explorer from Task Inference based Meta-Reinforcement Learning, SETIMRL). In this framework, an explorer agent is specially designed for task inference. The explorer takes the task exploration fully, and transits the collected data to the inference network. And the actor will adapt to the new tasks rapidly with the received inference information, which helps improve the model’s performance. Experimental results show that the proposed algorithm has better efficiency in multi-dimensions and sequential control tasks, compared to traditional meta reinforcement learning based on task inference.
传统的基于任务推理的元强化学习将任务推理与任务控制分离开来，忽视了任务推理过程中探索的重要性。智能体在任务探索过程和任务控制过程中使用相同的策略，导致任务推理效率较低。为了解决这一问题，本文提出了一种基于任务推理的元强化学习框架(SETIMRL, separation Explorer from task inference based meta - reinforcement learning)。在这个框架中，资源管理器代理是专门为任务推理而设计的。探索者充分进行任务探索，并将收集到的数据传输到推理网络。行动者可以根据接收到的推理信息快速适应新的任务，从而提高模型的性能。实验结果表明，与传统的基于任务推理的元强化学习相比，该算法在多维、序列控制任务中具有更好的效率。
{"title":"Separating Explorer for Task Inference Based Meta Reinforcement Learning Algorithm","authors":"Lanlan Gong, Xinghong Ling, Jiahui Lu, Jiaqin Zhou, Liang Xue","doi":"10.1109/INSAI54028.2021.00048","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00048","url":null,"abstract":"Traditional meta reinforcement learning based on task inference separates task inference with task control, but ignores the importance of exploration during task inference. The agent uses the same policy for both task exploration process and task control process, which leads to low task inference efficiency. To solve this problem, this paper proposes a task inference based meta reinforcement learning framework (Separating Explorer from Task Inference based Meta-Reinforcement Learning, SETIMRL). In this framework, an explorer agent is specially designed for task inference. The explorer takes the task exploration fully, and transits the collected data to the inference network. And the actor will adapt to the new tasks rapidly with the received inference information, which helps improve the model’s performance. Experimental results show that the proposed algorithm has better efficiency in multi-dimensions and sequential control tasks, compared to traditional meta reinforcement learning based on task inference.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123671991","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}
Current transformer is widely used in distribution network line protection, but due to its own structure limitation and measurement speed limitation, it will have a certain impact on the fault sampling quantity. This paper mainly analyzes the possible influence of current transformer, studies its tolerance to transition resistance and its influence on line selection protection, so as to provide a theoretical basis for distribution network fault identification and line protection.
{"title":"Research on the Influence of the Data Sampling from Current Transformer on Line Selection Protection of Grounding Faults","authors":"Zhi Li, Shaofeng Yu, Yong Li, Haijun Chen, Shibiao Tang, Wansheng Zhang","doi":"10.1109/INSAI54028.2021.00047","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00047","url":null,"abstract":"Current transformer is widely used in distribution network line protection, but due to its own structure limitation and measurement speed limitation, it will have a certain impact on the fault sampling quantity. This paper mainly analyzes the possible influence of current transformer, studies its tolerance to transition resistance and its influence on line selection protection, so as to provide a theoretical basis for distribution network fault identification and line protection.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123771936","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 : 2021-11-01DOI: 10.1109/INSAI54028.2021.00017
Zhigen Xu, Yanzhen Li, Liguo Shi, Weishuai Gong
In order to promote the development of intelligent, efficient and information-based ship power and power system, it is necessary to design a set of safe and effective, real-time monitoring, intelligent regulation and emergency reliable charging system for the ship, in order to charge the battery pack, electric propulsion system, lighting equipment, communication equipment, power supply system and so on The early warning system and the ship's daily electrical equipment carry out unified monitoring and intelligent management operation. In China, the shore power connection is mainly completed by manual towing of cables and manual docking of cable joints. This kind of operation requires the cooperation of terminal operators and mooring personnel. The operation is labor-intensive, inefficient and the working environment is relatively bad. Therefore, in order to solve the automation problem of shore power docking, improve work efficiency and intelligent level, and accelerate the promotion of demonstration application of shore power.
{"title":"Research on Test System of Ship Cable Charging and Docking","authors":"Zhigen Xu, Yanzhen Li, Liguo Shi, Weishuai Gong","doi":"10.1109/INSAI54028.2021.00017","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00017","url":null,"abstract":"In order to promote the development of intelligent, efficient and information-based ship power and power system, it is necessary to design a set of safe and effective, real-time monitoring, intelligent regulation and emergency reliable charging system for the ship, in order to charge the battery pack, electric propulsion system, lighting equipment, communication equipment, power supply system and so on The early warning system and the ship's daily electrical equipment carry out unified monitoring and intelligent management operation. In China, the shore power connection is mainly completed by manual towing of cables and manual docking of cable joints. This kind of operation requires the cooperation of terminal operators and mooring personnel. The operation is labor-intensive, inefficient and the working environment is relatively bad. Therefore, in order to solve the automation problem of shore power docking, improve work efficiency and intelligent level, and accelerate the promotion of demonstration application of shore power.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121434698","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 : 2021-11-01DOI: 10.1109/INSAI54028.2021.00026
Xiaoxiao Wang, Huiliang Shang, Linhua Jiang
Point clouds has been increasingly used in computer vision tasks like 3D reconstructions and robotic perceptions, and point cloud registration plays a key role in those scenarios. PPF (Point Pair Feature) based voting matching scheme is a widely-used method for point clouds registration. Based on actual experiences of using PPF we proposed modifications from 2 aspects. The first is based on point-pair visibility during offline dictionary generation and we applied 2 steps of spherical dictionary generation and merging instead of brute-force traversal. The second came from high-frequency information loss (richness of surface normal distribution decreasing) during point cloud downsampling, and we proposed a curvature-aware voxel downsampling method instead of uniform downsampling. We demonstrated the effectiveness of the proposals above via several experiments.
{"title":"Improved Point Pair Feature based Cloud Registration on Visibility and Downsampling","authors":"Xiaoxiao Wang, Huiliang Shang, Linhua Jiang","doi":"10.1109/INSAI54028.2021.00026","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00026","url":null,"abstract":"Point clouds has been increasingly used in computer vision tasks like 3D reconstructions and robotic perceptions, and point cloud registration plays a key role in those scenarios. PPF (Point Pair Feature) based voting matching scheme is a widely-used method for point clouds registration. Based on actual experiences of using PPF we proposed modifications from 2 aspects. The first is based on point-pair visibility during offline dictionary generation and we applied 2 steps of spherical dictionary generation and merging instead of brute-force traversal. The second came from high-frequency information loss (richness of surface normal distribution decreasing) during point cloud downsampling, and we proposed a curvature-aware voxel downsampling method instead of uniform downsampling. We demonstrated the effectiveness of the proposals above via several experiments.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"70 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127315640","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}
To realize the real-time flow measurement of hydrological stations, improve the accuracy of flow measurement at stations and reduce the intensity of manual flow measurement, the ADCP flow online monitoring system was introduced in Lanxi Hydrological Station in November 2020. After long-term monitoring and calibration analysis of a large amount of data, the model of correlation between the average flow velocity measured by ADCP and the average flow velocity of Lanxi section was established, which ensured that the station personnel could accurately grasp the flow characteristics and control conditions of the section, accurately calculate hourly flow and various characteristic values, capture the flood process of the section, and cover all flood periods. At the same time, it further explained that the specific gradient and flow velocity of Lanxi section changed correspondingly at stations which were frequently regulated by power stations and greatly affected by water conservancy projects, which directly affected the relationship between flow velocity, water level and flow, resulting in backwater jacking.
{"title":"Research on Real-time Flow Observation and Calculation Based on ADCP System","authors":"Dekang Zhu, Jianbin Guo, Xiang Cheng, Yanze Zhu, Shuqiao Fang, Feng Zhang","doi":"10.1109/INSAI54028.2021.00043","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00043","url":null,"abstract":"To realize the real-time flow measurement of hydrological stations, improve the accuracy of flow measurement at stations and reduce the intensity of manual flow measurement, the ADCP flow online monitoring system was introduced in Lanxi Hydrological Station in November 2020. After long-term monitoring and calibration analysis of a large amount of data, the model of correlation between the average flow velocity measured by ADCP and the average flow velocity of Lanxi section was established, which ensured that the station personnel could accurately grasp the flow characteristics and control conditions of the section, accurately calculate hourly flow and various characteristic values, capture the flood process of the section, and cover all flood periods. At the same time, it further explained that the specific gradient and flow velocity of Lanxi section changed correspondingly at stations which were frequently regulated by power stations and greatly affected by water conservancy projects, which directly affected the relationship between flow velocity, water level and flow, resulting in backwater jacking.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125852020","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 : 2021-11-01DOI: 10.1109/INSAI54028.2021.00053
Shenghui Wang, Jin Xu
Intelligent building is the inevitable outcome of information technology in today's world, and people's safety awareness is also improving with the development of society, and the safety issue has become an important proposition that everyone pays attention to. License plate number recognition technology is an important part of intelligent building security system, which plays an important role in the field of vehicle identification. This design is based on the deep learning method to realize the license plate recognition in the field of vehicle identification, including the preprocessing of license plate image, the location of license plate area, the segmentation of license plate characters and the recognition of license plate characters. The test shows that the license plate number recognition system designed this time can effectively and accurately recognize the license plate number in the image, and can effectively identify the identity.
{"title":"Application of License Plate Number Recognition Based on Deep Learning Method in Intelligent Building Security System","authors":"Shenghui Wang, Jin Xu","doi":"10.1109/INSAI54028.2021.00053","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00053","url":null,"abstract":"Intelligent building is the inevitable outcome of information technology in today's world, and people's safety awareness is also improving with the development of society, and the safety issue has become an important proposition that everyone pays attention to. License plate number recognition technology is an important part of intelligent building security system, which plays an important role in the field of vehicle identification. This design is based on the deep learning method to realize the license plate recognition in the field of vehicle identification, including the preprocessing of license plate image, the location of license plate area, the segmentation of license plate characters and the recognition of license plate characters. The test shows that the license plate number recognition system designed this time can effectively and accurately recognize the license plate number in the image, and can effectively identify the identity.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114895952","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 : 2021-11-01DOI: 10.1109/INSAI54028.2021.00065
Ling Hu, Lanlan Yin, Feng Mo
The graphics of the rose line algorithm are rich and diverse. This article introduces two methods of drawing the rose line algorithm. One of the main drawing methods is to use the C language EasyX graphics library to simulate the rose curve algorithm. Secondly, the author also use Python’s graphics library for drawing. By comparing the two methods, it found that the former is more capable of drawing complex rose line graphics. These drawn complex graphics can not only be used in the field of education, such as to increase students’ interest in programming, but also in the printing and dyeing industry to generate beautiful decorative patterns and graphics.
{"title":"Realization of Moving Graph Based on Rose Line Algorithm Using EasyX Method","authors":"Ling Hu, Lanlan Yin, Feng Mo","doi":"10.1109/INSAI54028.2021.00065","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00065","url":null,"abstract":"The graphics of the rose line algorithm are rich and diverse. This article introduces two methods of drawing the rose line algorithm. One of the main drawing methods is to use the C language EasyX graphics library to simulate the rose curve algorithm. Secondly, the author also use Python’s graphics library for drawing. By comparing the two methods, it found that the former is more capable of drawing complex rose line graphics. These drawn complex graphics can not only be used in the field of education, such as to increase students’ interest in programming, but also in the printing and dyeing industry to generate beautiful decorative patterns and graphics.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"122 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127068137","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 : 2021-11-01DOI: 10.1109/INSAI54028.2021.00058
Haoyu Wang, Chao Xu, Binzhong Wang, Runqiu Wu
The complex cosmic environment imposes communication challenges on space-ground cooperation, including high message delay, unstable communication link, and high bit error rate of spacecraft. It is of great importance to improve the communication quality between the mission center and the spacecraft, in particular with the control of remote space manipulator. In this paper, we design a set of message queue communication modules based on a mixture framework of TCP protocol and UDP protocol to improve the safety and efficiency of space-ground communication. By categorizing and re-ordering the common message commands, and sorting out the order importance, the topological sorting algorithm is used to optimize the transmission order of messages. In addition, the command sending mode is determined by the degree of network congestion. The simulation experiment shows that the proposed message queue design improves the general transmission safety. In particular, the transmission efficiency of emergency commands increased by about 30% in the case of insufficient network resources. This message queue designation scheme has potentials to be used in assisting space-ground cooperation.
{"title":"An Efficient Message Queue Designation Scheme for Space-Ground Communication","authors":"Haoyu Wang, Chao Xu, Binzhong Wang, Runqiu Wu","doi":"10.1109/INSAI54028.2021.00058","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00058","url":null,"abstract":"The complex cosmic environment imposes communication challenges on space-ground cooperation, including high message delay, unstable communication link, and high bit error rate of spacecraft. It is of great importance to improve the communication quality between the mission center and the spacecraft, in particular with the control of remote space manipulator. In this paper, we design a set of message queue communication modules based on a mixture framework of TCP protocol and UDP protocol to improve the safety and efficiency of space-ground communication. By categorizing and re-ordering the common message commands, and sorting out the order importance, the topological sorting algorithm is used to optimize the transmission order of messages. In addition, the command sending mode is determined by the degree of network congestion. The simulation experiment shows that the proposed message queue design improves the general transmission safety. In particular, the transmission efficiency of emergency commands increased by about 30% in the case of insufficient network resources. This message queue designation scheme has potentials to be used in assisting space-ground cooperation.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"76 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125621078","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}
Aiming at the requirement of collision-free and precise synchronized kinematics autonomous coordinated control of the dual-manipulator heavy-duty robot in carrying, grasping and dismantling operations, this paper studied the method of collision avoidance path planning in the dual-manipulator operation. The dual mechanical arm is an important part of the system, and its working state is directly related to the stability and reliability of the entire system. In this paper, we used the improved DH method to mathematically model the position and posture of the dual robotic arms. Under the robot operating system, we adopted Moveit! to control the movement towards the robotic arms. Finally, under obstacles at different distances, we simulated the improved fusion algorithm and compared it with the existing artificial potential field method and RRT* algorithm. The experimental results show that the improved fusion algorithm has better path planning effect and can make the robot arm reach the goal smoothly along the path of the least cost without collision.
{"title":"Collision Avoidance Path Planning of Nuclear Robot with Dual Manipulators","authors":"Xiangjun Liu, Chenyang Sun, Linjunhao Xiao, Runjie Shen","doi":"10.1109/INSAI54028.2021.00044","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00044","url":null,"abstract":"Aiming at the requirement of collision-free and precise synchronized kinematics autonomous coordinated control of the dual-manipulator heavy-duty robot in carrying, grasping and dismantling operations, this paper studied the method of collision avoidance path planning in the dual-manipulator operation. The dual mechanical arm is an important part of the system, and its working state is directly related to the stability and reliability of the entire system. In this paper, we used the improved DH method to mathematically model the position and posture of the dual robotic arms. Under the robot operating system, we adopted Moveit! to control the movement towards the robotic arms. Finally, under obstacles at different distances, we simulated the improved fusion algorithm and compared it with the existing artificial potential field method and RRT* algorithm. The experimental results show that the improved fusion algorithm has better path planning effect and can make the robot arm reach the goal smoothly along the path of the least cost without collision.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125660558","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 : 2021-11-01DOI: 10.1109/INSAI54028.2021.00035
Jing Liu, Y. Liu, Jieyu Lin, Donglai Wei, Xu Xia, Wei Ni, Xiaohong Huang, Liang Song
Abnormal driving behavior detection (ADBD) is essential to improving driving safety. Traditional approaches usually extract features manually, resulting in insufficient exploration of deep features of driving behavior. To address the limitation of shallow feature-based approaches, we propose a one-dimensional convolutional neural network (1D CNN) model for ADBD and transforms it into the time series multi-classification. Firstly, we construct a dataset of driving behavior using smartphone sensors, and five fine-grained abnormal driving behaviors (Hard braking, Weaving, Swerving, Quick turn, Quick U-turn) are defined and labeled by analyzing the abnormal driving feature patterns. Then, we need to map the smartphone coordinate to the vehicle coordinate, and apply the low-pass filtering on the input data for high-frequency noise reduction. Finally, we train the dataset with the 1D CNN model for feature extraction and classification. The experimental results show that the proposed 1D CNN model efficiently achieves multi-classification of abnormal driving behaviors with an average accuracy of 97%, significantly better than the traditional algorithm of k-nearest neighbor and support vector machine.
{"title":"One-Dimensional Convolutional Neural Network Model for Abnormal Driving Behaviors Detection Using Smartphone Sensors","authors":"Jing Liu, Y. Liu, Jieyu Lin, Donglai Wei, Xu Xia, Wei Ni, Xiaohong Huang, Liang Song","doi":"10.1109/INSAI54028.2021.00035","DOIUrl":"https://doi.org/10.1109/INSAI54028.2021.00035","url":null,"abstract":"Abnormal driving behavior detection (ADBD) is essential to improving driving safety. Traditional approaches usually extract features manually, resulting in insufficient exploration of deep features of driving behavior. To address the limitation of shallow feature-based approaches, we propose a one-dimensional convolutional neural network (1D CNN) model for ADBD and transforms it into the time series multi-classification. Firstly, we construct a dataset of driving behavior using smartphone sensors, and five fine-grained abnormal driving behaviors (Hard braking, Weaving, Swerving, Quick turn, Quick U-turn) are defined and labeled by analyzing the abnormal driving feature patterns. Then, we need to map the smartphone coordinate to the vehicle coordinate, and apply the low-pass filtering on the input data for high-frequency noise reduction. Finally, we train the dataset with the 1D CNN model for feature extraction and classification. The experimental results show that the proposed 1D CNN model efficiently achieves multi-classification of abnormal driving behaviors with an average accuracy of 97%, significantly better than the traditional algorithm of k-nearest neighbor and support vector machine.","PeriodicalId":232335,"journal":{"name":"2021 International Conference on Networking Systems of AI (INSAI)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133056527","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：