Pub Date : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990122
Tien Nguyen, Nam-Cuong Nguyen, Duy-Khanh Ngo, Viet-Lam Phan, Minh-Hung Pham, Duc-An Nguyen, Minh-Hiep Doan, Thi-Lan Le
While isolated hand gesture recognition methods aims to determine the type of gestures for a given sequence, continuous hand gesture recognition methods have to perform one more task: determining the starting point and ending point of the hand gesture. This task becomes challenging as the starting point and ending points of the gestures are not usually obvious even for human being. This paper presents a method for continuous hand gesture recognition based on skeleton information that consists of two phases: gesture detection and gesture recognition. In our method, to leverage the lightweight and the robustness of recognition models, TD-Net (Triple Feature Double Motion) model is employed in both gesture detection and recognition phases. Experimental results on IPN dataset have shown that the proposed method outperforms different state-of-the-art methods with 40.10% of Levenshtein accuracy and 0.1ms of inference time.
{"title":"A Continuous Real-time Hand Gesture Recognition Method based on Skeleton","authors":"Tien Nguyen, Nam-Cuong Nguyen, Duy-Khanh Ngo, Viet-Lam Phan, Minh-Hung Pham, Duc-An Nguyen, Minh-Hiep Doan, Thi-Lan Le","doi":"10.1109/ICCAIS56082.2022.9990122","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990122","url":null,"abstract":"While isolated hand gesture recognition methods aims to determine the type of gestures for a given sequence, continuous hand gesture recognition methods have to perform one more task: determining the starting point and ending point of the hand gesture. This task becomes challenging as the starting point and ending points of the gestures are not usually obvious even for human being. This paper presents a method for continuous hand gesture recognition based on skeleton information that consists of two phases: gesture detection and gesture recognition. In our method, to leverage the lightweight and the robustness of recognition models, TD-Net (Triple Feature Double Motion) model is employed in both gesture detection and recognition phases. Experimental results on IPN dataset have shown that the proposed method outperforms different state-of-the-art methods with 40.10% of Levenshtein accuracy and 0.1ms of inference time.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115723818","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990036
T. Trong, Manh Vu Van, Quan-Tran Hai, B. N. Thai
We propose an autonomous landing scheme for Vertical Take-off and Landing Unmanned Aerial Vehicle (VTOL UAV) on a moving ship at sea and this scheme is embedded in a hardware platform - Companion Computer. This mission requires determining the ship’s location, speed, and trajectory, which are significant challenges in the marine environment. This research applies a non-contact method, it is combined deep-learning and visual servoing techniques for real-time measuring of the parameters just mentioned above and tightly coupling with modern navigation logic to ensure the UAV follows a fast and optimal landing trajectory. No prior information about a moving ship’s location and landing pad is needed during the entire VTOL UAV’s landing process. The method aims to improve the performance of the landing. The proposed technique has been evaluated in a hardware in the loop simulation system using Jetson Nano and X-Plane.
{"title":"Autonomous Landing Scheme of VTOL UAV on Moving Ship Using Deep Learning Technique Embedded in Companion Computer","authors":"T. Trong, Manh Vu Van, Quan-Tran Hai, B. N. Thai","doi":"10.1109/ICCAIS56082.2022.9990036","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990036","url":null,"abstract":"We propose an autonomous landing scheme for Vertical Take-off and Landing Unmanned Aerial Vehicle (VTOL UAV) on a moving ship at sea and this scheme is embedded in a hardware platform - Companion Computer. This mission requires determining the ship’s location, speed, and trajectory, which are significant challenges in the marine environment. This research applies a non-contact method, it is combined deep-learning and visual servoing techniques for real-time measuring of the parameters just mentioned above and tightly coupling with modern navigation logic to ensure the UAV follows a fast and optimal landing trajectory. No prior information about a moving ship’s location and landing pad is needed during the entire VTOL UAV’s landing process. The method aims to improve the performance of the landing. The proposed technique has been evaluated in a hardware in the loop simulation system using Jetson Nano and X-Plane.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131679584","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990461
Nguyen Xuan Phuc, N. Tung, Truong Hoang Nam
Nowadays, Vietnam's power system will develop an HVDC system quickly to respond to the national energy transmission demand. The HVDC point-to-point link is being met on demand for preliminary and final designs. For the Vietnamese grid system, the procedure for HVDC, essentials of insulation coordination, are necessary, and transient overvoltage is considered. Transient Over-voltages values are conducted by numerical simulation. Parametric studies using PSSE calculations and transfer in EMTP-ATP software are shown in this paper to simulate the worst-case fault locations inside the converter station to have the maximum value possibility of transient overvoltage. The parametric of the Vietnamese power system and the MMC-HVDC model are provided and analyzed. The results offer insights for engineers involved in the insulation coordination of the MMC-HVDC link.
{"title":"Analyze the Transient Overvoltages in the station of Vietnamese model HVDC-MMC system","authors":"Nguyen Xuan Phuc, N. Tung, Truong Hoang Nam","doi":"10.1109/ICCAIS56082.2022.9990461","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990461","url":null,"abstract":"Nowadays, Vietnam's power system will develop an HVDC system quickly to respond to the national energy transmission demand. The HVDC point-to-point link is being met on demand for preliminary and final designs. For the Vietnamese grid system, the procedure for HVDC, essentials of insulation coordination, are necessary, and transient overvoltage is considered. Transient Over-voltages values are conducted by numerical simulation. Parametric studies using PSSE calculations and transfer in EMTP-ATP software are shown in this paper to simulate the worst-case fault locations inside the converter station to have the maximum value possibility of transient overvoltage. The parametric of the Vietnamese power system and the MMC-HVDC model are provided and analyzed. The results offer insights for engineers involved in the insulation coordination of the MMC-HVDC link.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115121476","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990130
Yiru Lin, Yuanhang Wu, Wei Yi
Maritime radar target detection is often affected by sea clutter, and the detection performance in the case of low signal-to-clutter ratio (SCR) is usually poor. In this paper, we propose a two-stage deep learning method for sea clutter suppression and point target detection. Take the cluttered Range-Doppler (RD) spectra as input, at the first stage, reconstructed RD spectra are obtained as clutter suppression results through Attention Denoising Adversarial-Autoencoders (Atten-DAAE). At the second stage, detection results are obtained through the traditional one-stage detection network YOLOv5s. The proposed method has been verified on two datasets with simulated and measured clutter data respectively and compared with the traditional method and other networks, which shows better detection performance.
{"title":"Two-stage Networks with Adversarial Clutter Suppression for Maritime Radar Target Detection","authors":"Yiru Lin, Yuanhang Wu, Wei Yi","doi":"10.1109/ICCAIS56082.2022.9990130","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990130","url":null,"abstract":"Maritime radar target detection is often affected by sea clutter, and the detection performance in the case of low signal-to-clutter ratio (SCR) is usually poor. In this paper, we propose a two-stage deep learning method for sea clutter suppression and point target detection. Take the cluttered Range-Doppler (RD) spectra as input, at the first stage, reconstructed RD spectra are obtained as clutter suppression results through Attention Denoising Adversarial-Autoencoders (Atten-DAAE). At the second stage, detection results are obtained through the traditional one-stage detection network YOLOv5s. The proposed method has been verified on two datasets with simulated and measured clutter data respectively and compared with the traditional method and other networks, which shows better detection performance.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129876526","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990491
Vo Thi Thanh Ha
In an attempt to decrease the torque ripple of the SRM motor in an electric car application, the direct torque control (DTC) approach, based on the fuzzy logic controller (FLC), is described in this research. The fuzzy logic controller has two inputs: speed and speed error, which are utilized by the SRM torque control loop. The 49-rule set also controls the FLC. The results show that the FLC reduces the motor's torque ripple more effectively than PI controllers. It makes the drive insensitive to parameter changes and compensates for the nonlinear torque characteristics of SRM. Additionally, the electromagnetic torque of the proposed FLC is more uniform than that of the conventional PI. The effectiveness of this FLC has also been demonstrated using MATLAB/SIMULINK simulation. Through simulation, the significance of this FLC technique has been shown.
{"title":"Torque Ripple Reduction of the SRM Motor Using Nonlinear Controller for Electric Vehicles Application","authors":"Vo Thi Thanh Ha","doi":"10.1109/ICCAIS56082.2022.9990491","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990491","url":null,"abstract":"In an attempt to decrease the torque ripple of the SRM motor in an electric car application, the direct torque control (DTC) approach, based on the fuzzy logic controller (FLC), is described in this research. The fuzzy logic controller has two inputs: speed and speed error, which are utilized by the SRM torque control loop. The 49-rule set also controls the FLC. The results show that the FLC reduces the motor's torque ripple more effectively than PI controllers. It makes the drive insensitive to parameter changes and compensates for the nonlinear torque characteristics of SRM. Additionally, the electromagnetic torque of the proposed FLC is more uniform than that of the conventional PI. The effectiveness of this FLC has also been demonstrated using MATLAB/SIMULINK simulation. Through simulation, the significance of this FLC technique has been shown.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125337405","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990376
Qi Li, Pengchao Tian, Ye Shi, Yuanming Shi, H. Tuan
Recent years have seen the wide applications of renewable energy sources and plug-in electric vehicles in smart grids. However, their inherent uncertainties may lead to serious voltage deviations, load fluctuations and power losses. In this paper, we formulate a distributionally robust optimization (DRO) for vehicle-to-grid considering the uncertainties of solar power and PEVs. We utilize conditional value at risk to quantify the risk of violating inequalities containing uncertainties and the Wasserstein metric to reformulate the DRO problem into a tractable convex optimization problem. The DRO is implemented under a model predictive control framework to further reduce the uncertainties of PEVs and RESs. Numerical experiment results validate the efficiency of our method.
{"title":"Distributionally Robust Optimization for Vehicle-to-grid with Uncertain Renewable Energy","authors":"Qi Li, Pengchao Tian, Ye Shi, Yuanming Shi, H. Tuan","doi":"10.1109/ICCAIS56082.2022.9990376","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990376","url":null,"abstract":"Recent years have seen the wide applications of renewable energy sources and plug-in electric vehicles in smart grids. However, their inherent uncertainties may lead to serious voltage deviations, load fluctuations and power losses. In this paper, we formulate a distributionally robust optimization (DRO) for vehicle-to-grid considering the uncertainties of solar power and PEVs. We utilize conditional value at risk to quantify the risk of violating inequalities containing uncertainties and the Wasserstein metric to reformulate the DRO problem into a tractable convex optimization problem. The DRO is implemented under a model predictive control framework to further reduce the uncertainties of PEVs and RESs. Numerical experiment results validate the efficiency of our method.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128714380","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990471
Duy-Dinh Nguyen, The-Tiep Pham, T. Le
Dual-Active-Bridge (DAB)-liked converters owning the inherent soft-switching capability, hence, they can achieve very high system efficiency. In order for soft-switching to achieve, the body diode of a MOSFET should conduct before the switch is triggered. This diode conduction occurs in the dead-time. If the dead-time is large, diode conduction loss may be significant that leads to the downgrade of system performance. This unwanted loss can be eliminated by triggering the MOSFET right at the instant when the transition completed. This paper proposed a method to detect that instant By using several low-cost passive components, a transition detection circuit can be built which can detect both the tura-on and turn-off transition completions. The effectiveness of the circuit is verified by simulation for DAB converter applications. However, it can also be used for detecting transition of other bridge-typed power electronic converter topologies.
{"title":"A Transition Completion Detection Circuit for Dual Active Bridge Converters","authors":"Duy-Dinh Nguyen, The-Tiep Pham, T. Le","doi":"10.1109/ICCAIS56082.2022.9990471","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990471","url":null,"abstract":"Dual-Active-Bridge (DAB)-liked converters owning the inherent soft-switching capability, hence, they can achieve very high system efficiency. In order for soft-switching to achieve, the body diode of a MOSFET should conduct before the switch is triggered. This diode conduction occurs in the dead-time. If the dead-time is large, diode conduction loss may be significant that leads to the downgrade of system performance. This unwanted loss can be eliminated by triggering the MOSFET right at the instant when the transition completed. This paper proposed a method to detect that instant By using several low-cost passive components, a transition detection circuit can be built which can detect both the tura-on and turn-off transition completions. The effectiveness of the circuit is verified by simulation for DAB converter applications. However, it can also be used for detecting transition of other bridge-typed power electronic converter topologies.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127535217","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990197
Anh Tan Nguyen, D. Nguyen
The second-order generalized integrator (SOGI) has been receiving a great deal of attention recently for the grid voltage estimation owing to its excellent harmonic rejection capability. In practice, such an algorithm is usually implemented by digital devices, thus the accurate discretization of the SOGI is essential to achieve the desired system performance. A comprehensive study about the discretization of the SOGI for grid voltage estimation is presented in this paper, in which the performance comparisons of SOGI discretized by various methods, i.e., backward Euler method, Tustin method, ZOH method, impulse invariant method, are conducted. Through the thorough analysis and simulation results, the proper discretization method for the SOGI for grid voltage estimation has been recommended.
{"title":"Implementation of Discretization Methods for Second-Order Generalized Integrator in Grid Voltage Estimation Systems","authors":"Anh Tan Nguyen, D. Nguyen","doi":"10.1109/ICCAIS56082.2022.9990197","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990197","url":null,"abstract":"The second-order generalized integrator (SOGI) has been receiving a great deal of attention recently for the grid voltage estimation owing to its excellent harmonic rejection capability. In practice, such an algorithm is usually implemented by digital devices, thus the accurate discretization of the SOGI is essential to achieve the desired system performance. A comprehensive study about the discretization of the SOGI for grid voltage estimation is presented in this paper, in which the performance comparisons of SOGI discretized by various methods, i.e., backward Euler method, Tustin method, ZOH method, impulse invariant method, are conducted. Through the thorough analysis and simulation results, the proper discretization method for the SOGI for grid voltage estimation has been recommended.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127823026","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990480
Walter Brescia, L. D. Cicco, S. Mascolo
Reinforcement Learning (RL) has gained interest in the control and automation communities thanks to its encouraging results in many challenging control problems without requiring a model of the system and of the environment. Yet, it is well-known that employing such a learning-based approach in real scenarios may be problematic, as a prohibitive amount of data might be required to converge to an optimal control policy. In this work, we equip a popular RL algorithm with two tools to improve exploration effectiveness and sample efficiency: the Episodic Noise, that helps useful subsets of actions emerge already in the first few training episodes, and the Difficulty Manager, that generates goals proportioned to the current agent’s capabilities. We demonstrate the effectiveness of such proposed tools on a pose regulation task of a four wheel steering four wheel driving robot, suitable for a wide range of industrial scenarios. The resulting agent learns effective sets of actions in just a few hundreds training epochs, reaching satisfactory performance during tests.
{"title":"Sample-Efficient Reinforcement Learning for Pose Regulation of a Mobile Robot","authors":"Walter Brescia, L. D. Cicco, S. Mascolo","doi":"10.1109/ICCAIS56082.2022.9990480","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990480","url":null,"abstract":"Reinforcement Learning (RL) has gained interest in the control and automation communities thanks to its encouraging results in many challenging control problems without requiring a model of the system and of the environment. Yet, it is well-known that employing such a learning-based approach in real scenarios may be problematic, as a prohibitive amount of data might be required to converge to an optimal control policy. In this work, we equip a popular RL algorithm with two tools to improve exploration effectiveness and sample efficiency: the Episodic Noise, that helps useful subsets of actions emerge already in the first few training episodes, and the Difficulty Manager, that generates goals proportioned to the current agent’s capabilities. We demonstrate the effectiveness of such proposed tools on a pose regulation task of a four wheel steering four wheel driving robot, suitable for a wide range of industrial scenarios. The resulting agent learns effective sets of actions in just a few hundreds training epochs, reaching satisfactory performance during tests.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"212 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115785460","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 : 2022-11-21DOI: 10.1109/ICCAIS56082.2022.9990287
A. Nguyen, Duc Minh Nguyen, V. Pham, H. Nguyen, D. T. Tran, J.-H. Lee, A. Q. Nguyen
Localization or state estimation is one of the most important tasks for UAVs based on different kinds of sensors such as GPS, IMU, Lidar or cameras. However, localization based on only a monocular camera or visual odometry is one of the most challenging research topics. Conventional methods are proposed based on the detection of key features in each image and matching them on consecutive images to estimate the camera motions. Deep-learning methods have also been studied to solve the problem. Although the current learning-based visual odometry methods score high results on public datasets, there is a lack of real-time implementation of the methods in common robot operating systems such as ROS to integrate them into a navigation system. In this paper, we introduce a ROS implementation of state-of-the-art conventional feature-based method, ORB-SLAM3, together with a deep-learning-based method, SC-SfMLearner for real-time UAV localization. A photo-realistic simulator, Flightmare, is used to test the implementation together with another navigation task such as control. The implementation can evaluate both algorithms in real-time operation to compare their performances. Based on evaluation results from the simulated environments, the limitation or failure cases of the algorithms could be found, then, the best parameters of the algorithms can be adjusted to improve the algorithms to avoid failures in practical experiments.
{"title":"Real-time ROS Implementation of Conventional Feature-based and Deep-learning-based Monocular Visual Odometry for UAV","authors":"A. Nguyen, Duc Minh Nguyen, V. Pham, H. Nguyen, D. T. Tran, J.-H. Lee, A. Q. Nguyen","doi":"10.1109/ICCAIS56082.2022.9990287","DOIUrl":"https://doi.org/10.1109/ICCAIS56082.2022.9990287","url":null,"abstract":"Localization or state estimation is one of the most important tasks for UAVs based on different kinds of sensors such as GPS, IMU, Lidar or cameras. However, localization based on only a monocular camera or visual odometry is one of the most challenging research topics. Conventional methods are proposed based on the detection of key features in each image and matching them on consecutive images to estimate the camera motions. Deep-learning methods have also been studied to solve the problem. Although the current learning-based visual odometry methods score high results on public datasets, there is a lack of real-time implementation of the methods in common robot operating systems such as ROS to integrate them into a navigation system. In this paper, we introduce a ROS implementation of state-of-the-art conventional feature-based method, ORB-SLAM3, together with a deep-learning-based method, SC-SfMLearner for real-time UAV localization. A photo-realistic simulator, Flightmare, is used to test the implementation together with another navigation task such as control. The implementation can evaluate both algorithms in real-time operation to compare their performances. Based on evaluation results from the simulated environments, the limitation or failure cases of the algorithms could be found, then, the best parameters of the algorithms can be adjusted to improve the algorithms to avoid failures in practical experiments.","PeriodicalId":273404,"journal":{"name":"2022 11th International Conference on Control, Automation and Information Sciences (ICCAIS)","volume":"149 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124166274","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：