scholarly journals Real-Time Closed-Loop Detection Method of vSLAM Based on a Dynamic Siamese Network

Sensors ◽  
2021 ◽  
Vol 21 (22) ◽  
pp. 7612
Author(s):  
Quande Yuan ◽  
Zhenming Zhang ◽  
Yuzhen Pi ◽  
Lei Kou ◽  
Fangfang Zhang
Keyword(s):  
Real Time ◽  
Closed Loop ◽  
Detection Method ◽  
Model Parameters ◽  
Training Strategy ◽  
Key Frame ◽  
Siamese Network ◽  
Localization And Mapping ◽  
Loop Detection ◽  
The Impact

As visual simultaneous localization and mapping (vSLAM) is easy disturbed by the changes of camera viewpoint and scene appearance when building a globally consistent map, the robustness and real-time performance of key frame image selections cannot meet the requirements. To solve this problem, a real-time closed-loop detection method based on a dynamic Siamese networks is proposed in this paper. First, a dynamic Siamese network-based fast conversion learning model is constructed to handle the impact of external changes on key frame judgments, and an elementwise convergence strategy is adopted to ensure the accurate positioning of key frames in the closed-loop judgment process. Second, a joint training strategy is designed to ensure the model parameters can be learned offline in parallel from tagged video sequences, which can effectively improve the speed of closed-loop detection. Finally, the proposed method is applied experimentally to three typical closed-loop detection scenario datasets and the experimental results demonstrate the effectiveness and robustness of the proposed method under the interference of complex scenes.

scholarly journals Comment on Can assimilation of crowdsourced data in hydrological modelling improve flood prediction? by Mazzoleni et al. (2017)

2017 ◽  
Author(s):  
Daniele P. Viero
Keyword(s):  
Real Time ◽  
Data Uncertainty ◽  
Model Parameters ◽  
Design Strategies ◽  
Actual System ◽  
Recent Contribution ◽  
Flood Prediction ◽  
Irregular Frequency ◽  
Crowdsourced Data ◽  
The Impact

Abstract. In their recent contribution, Mazzoleni et al. (2017) investigated the integration of crowdsourced data (CSD) in hydrological models to improve the accuracy of real-time flood forecast. They showed that assimilation of CSD improves the overall model performance in all the considered case studies. The impact of irregular frequency of available crowdsourced data, and that of data uncertainty, were also deeply assessed. However, it has to be remarked that, in their work, the Authors used synthetic (i.e., not actually measured) crowdsourced data, because actual crowdsourced data were not available at the moment of the study. This point, briefly mentioned by the authors, deserves further discussion. In most real-world applications, rainfall-runoff models are calibrated using data from traditional sensors. Typically, CSD are collected at different locations, where semi-distributed models are not calibrated. In a context of equifinality and of poor identifiability of model parameters, the model internal states can hardly mimic the actual system states away from calibration points, thus reducing the chances of success in assimilating real (i.e., not synthetic) CSD. Additional criteria are given that are useful for the a-priori evaluation of crowdsourced data for real-time flood forecasting and, hopefully, to plan apt design strategies for both model calibration and collection of crowdsourced data.

An Analysis on the Optimization of Closed-Loop Detection Method for Optical Voltage Sensor Based on Pockels Effect

2014 ◽  
Vol 32 (5) ◽  
pp. 1006-1013 ◽  
Author(s):  
Hui Li ◽  
Liyang Cui ◽  
Zhili Lin ◽  
Lijing Li ◽  
Chunxi Zhang
Keyword(s):  
Closed Loop ◽  
Detection Method ◽  
Voltage Sensor ◽  
Pockels Effect ◽  
Loop Detection

scholarly journals Real-Time RGB-D Simultaneous Localization and Mapping Guided by Terrestrial LiDAR Point Cloud for Indoor 3-D Reconstruction and Camera Pose Estimation

2019 ◽  
Vol 9 (16) ◽  
pp. 3264 ◽  
Author(s):  
Xujie Kang ◽  
Jing Li ◽  
Xiangtao Fan ◽  
Wenhui Wan
Keyword(s):  
Real Time ◽  
Point Cloud ◽  
Closed Loop ◽  
Simultaneous Localization And Mapping ◽  
Motion Blur ◽  
Long Distance ◽  
Terrestrial Lidar ◽  
Map Construction ◽  
Localization And Mapping ◽  
Camera Pose

In recent years, low-cost and lightweight RGB and depth (RGB-D) sensors, such as Microsoft Kinect, have made available rich image and depth data, making them very popular in the field of simultaneous localization and mapping (SLAM), which has been increasingly used in robotics, self-driving vehicles, and augmented reality. The RGB-D SLAM constructs 3D environmental models of natural landscapes while simultaneously estimating camera poses. However, in highly variable illumination and motion blur environments, long-distance tracking can result in large cumulative errors and scale shifts. To address this problem in actual applications, in this study, we propose a novel multithreaded RGB-D SLAM framework that incorporates a highly accurate prior terrestrial Light Detection and Ranging (LiDAR) point cloud, which can mitigate cumulative errors and improve the system’s robustness in large-scale and challenging scenarios. First, we employed deep learning to achieve system automatic initialization and motion recovery when tracking is lost. Next, we used terrestrial LiDAR point cloud to obtain prior data of the landscape, and then we applied the point-to-surface inductively coupled plasma (ICP) iterative algorithm to realize accurate camera pose control from the previously obtained LiDAR point cloud data, and finally expanded its control range in the local map construction. Furthermore, an innovative double window segment-based map optimization method is proposed to ensure consistency, better real-time performance, and high accuracy of map construction. The proposed method was tested for long-distance tracking and closed-loop in two different large indoor scenarios. The experimental results indicated that the standard deviation of the 3D map construction is 10 cm in a mapping distance of 100 m, compared with the LiDAR ground truth. Further, the relative cumulative error of the camera in closed-loop experiments is 0.09%, which is twice less than that of the typical SLAM algorithm (3.4%). Therefore, the proposed method was demonstrated to be more robust than the ORB-SLAM2 algorithm in complex indoor environments.

scholarly journals IMU-Assisted 2D SLAM Method for Low-Texture and Dynamic Environments

2018 ◽  
Vol 8 (12) ◽  
pp. 2534 ◽  
Author(s):  
Zhongli Wang ◽  
Yan Chen ◽  
Yue Mei ◽  
Kuo Yang ◽  
Baigen Cai
Keyword(s):  
Closed Loop ◽  
Moving Objects ◽  
Interpolation Method ◽  
Dynamic Environments ◽  
Open Loop ◽  
Indoor Environments ◽  
Matching Problem ◽  
Scan Matching ◽  
Localization And Mapping ◽  
Loop Detection

Generally, the key issues of 2D LiDAR-based simultaneous localization and mapping (SLAM) for indoor application include data association (DA) and closed-loop detection. Particularly, a low-texture environment, which refers to no obvious changes between two consecutive scanning outputs, with moving objects existing in the environment will bring great challenges on DA and the closed-loop detection, and the accuracy and consistency of SLAM may be badly affected. There is not much literature that addresses this issue. In this paper, a mapping strategy is firstly exploited to improve the performance of the 2D SLAM in dynamic environments. Secondly, a fusion method which combines the IMU sensor with a 2D LiDAR, based on framework of extended Kalman Filter (EKF), is proposed to enhance the performance under low-texture environments. In the front-end of the proposed SLAM method, initial motion estimation is obtained from the output of EKF, and it can be taken as the initial pose for the scan matching problem. Then the scan matching problem can be optimized by the Levenberg–Marquardt (LM) algorithm. For the back-end optimization, a sparse pose adjustment (SPA) method is employed. To improve the accuracy, the grid map is updated with the bicubic interpolation method for derivative computing. With the improvements both in the DA process and the back-end optimization stage, the accuracy and consistency of SLAM results in low-texture environments is enhanced. Qualitative and quantitative experiments with open-loop and closed-loop cases have been conducted and the results are analyzed, confirming that the proposed method is effective in low-texture and dynamic indoor environments.

Some Practical Issues on Real-Time Reservoir Model Updating Using Ensemble Kalman Filter

SPE Journal ◽  
2007 ◽  
Vol 12 (02) ◽  
pp. 156-166 ◽  
Author(s):  
Xian-Huan Wen ◽  
Wen H. Chen
Keyword(s):  
Real Time ◽  
Ensemble Kalman Filter ◽  
Closed Loop ◽  
Model Updating ◽  
Reservoir Management ◽  
Time Interval ◽  
Ensemble Size ◽  
Reservoir Model ◽  
Reservoir Models ◽  
The Impact

Summary The concept of "closed-loop" reservoir management is currently receiving considerable attention in the petroleum industry. A "real-time" or "continuous" reservoir model updating technique is a critical component for the feasible application of any closed-loop, model-based reservoir management process. This technique should be able to rapidly and continuously update reservoir models assimilating the up-to-date observations of production data so that the performance predictions and the associated uncertainty are up-to-date for optimization of future development/operations. The ensemble Kalman filter (EnKF) method has been shown to be quite efficient for this purpose in large-scale nonlinear systems. Previous studies show that a relatively large ensemble size is required for EnKF to reliably assess the uncertainty, and a confirming step is recommended to ensure the consistency of the updated static and dynamic variables with the flow equations. In this paper, we further explore the capability of EnKF, focusing on some practical issues including the correction of the linear and Gaussian assumptions during filter updating with iteration, the reduction of ensemble size with a resampling scheme, and the impact of data assimilation time interval. Results from the example in this paper demonstrate that the proposed iterative EnKF performs better with more accurate predictions and less uncertainty than the traditional noniterative EnKF. The use of iteration reduces the impact of nonlinearity and non-Gaussianity. Results also show that iteration may only be required when predictions are considerably deviated from the observations. The proposed resampling scheme can significantly reduce the ensemble size necessary for reliable assessment of uncertainty with improved accuracy. Finally, we show that the noniterative EnKF is sensitive to the size of time interval between the assimilation steps. Using the proposed iterative EnKF, results are more stable, more accurate reservoir models and predictions can be obtained even when a large time interval is used. This also indicates that iteration within the EnKF updating serves as a process that corrects the stronger nonlinear and non-Gaussian behaviors when larger time interval is used. Introduction Reservoir models have become an important part of day-to-day decision analysis related to management of oil/gas fields. The closed-loop reservoir management concept (Jansen et al. 2005) allows real-time decisions to be made that maximize the production potential of a reservoir. These decisions are based on the most current information available about the reservoir model and the associated uncertainty of the information. One critical requirement in this real-time, model-based reservoir management process is the ability to rapidly estimate the reservoir models and the associated uncertainty reflecting the most current production data in a real-time fashion. Based on a number of studies, the EnKF method was shown to be well-suited for such applications compared to the traditional history-matching (HM) methods (Evensen 1999; Gu and Oliver 2006; Wen and Chen 2006).

scholarly journals Road Artery Traffic Light Optimization with Use of the Reinforcement Learning

2014 ◽  
Vol 26 (2) ◽  
pp. 101-108 ◽  
Author(s):  
Rok Marsetič ◽  
Darja Šemrov ◽  
Marijan Žura
Keyword(s):  
Real Time ◽  
Learning Algorithm ◽  
Heuristic Method ◽  
Model Parameters ◽  
Traffic Light ◽  
Q Learning ◽  
Traffic Lights ◽  
Algorithm Comparison ◽  
Actuated Signal ◽  
The Impact

The basic principle of optimal traffic control is the appropriate real-time response to dynamic traffic flow changes. Signal plan efficiency depends on a large number of input parameters. An actuated signal system can adjust very well to traffic conditions, but cannot fully adjust to stochastic traffic volume oscillation. Due to the complexity of the problem analytical methods are not applicable for use in real time, therefore the purpose of this paper is to introduce heuristic method suitable for traffic light optimization in real time. With the evolution of artificial intelligence new possibilities for solving complex problems have been introduced. The goal of this paper is to demonstrate that the use of the Q learning algorithm for traffic lights optimization is suitable. The Q learning algorithm was verified on a road artery with three intersections. For estimation of the effectiveness and efficiency of the proposed algorithm comparison with an actuated signal plan was carried out. The results (average delay per vehicle and the number of vehicles that left road network) show that Q learning algorithm outperforms the actuated signal controllers. The proposed algorithm converges to the minimal delay per vehicle regardless of the stochastic nature of traffic. In this research the impact of the model parameters (learning rate, exploration rate, influence of communication between agents and reward type) on algorithm effectiveness were analysed as well.

scholarly journals Influence of Wake Model Superposition and Secondary Steering on Model-Based Wake Steering Control with SCADA Data Assimilation

Energies ◽  
2020 ◽  
Vol 14 (1) ◽  
pp. 52
Author(s):  
Michael F. Howland ◽  
John O. Dabiri
Keyword(s):  
Wind Farm ◽  
Closed Loop ◽  
Power Production ◽  
Model Parameters ◽  
Closed Loop Control ◽  
Steering Control ◽  
Loop Control ◽  
Large Eddy ◽  
Wake Model ◽  
The Impact

Methods for wind farm power optimization through the use of wake steering often rely on engineering wake models due to the computational complexity associated with resolving wind farm dynamics numerically. Within the transient, turbulent atmospheric boundary layer, closed-loop control is required to dynamically adjust to evolving wind conditions, wherein the optimal wake model parameters are estimated as a function of time in a hybrid physics- and data-driven approach using supervisory control and data acquisition (SCADA) data. Analytic wake models rely on wake velocity deficit superposition methods to generalize the individual wake deficit to collective wind farm flow. In this study, the impact of the wake model superposition methodologies on closed-loop control are tested in large eddy simulations of the conventionally neutral atmospheric boundary layer with full Coriolis effects. A model for the non-vanishing lateral velocity trailing a yaw misaligned turbine, termed secondary steering, is also presented, validated, and tested in the closed-loop control framework. Modified linear and momentum conserving wake superposition methodologies increase the power production in closed-loop wake steering control statistically significantly more than linear superposition. While the secondary steering model increases the power production and reduces the predictive error associated with the wake model, the impact is not statistically significant. Modified linear and momentum conserving superposition using the proposed secondary steering model increase a six turbine array power production, compared to baseline control, in large eddy simulations by 7.5% and 7.7%, respectively, with wake model predictive mean absolute errors of 0.03P1 and 0.04P1, respectively, where P1 is the baseline power production of the leading turbine in the array. Ensemble Kalman filter parameter estimation significantly reduces the wake model predictive error for all wake deficit superposition and secondary steering cases compared to predefined model parameters.

The Impact of Visual Feedback Type on the Mastery of Visuo-Motor Transformations

2012 ◽  
Vol 220 (1) ◽  
pp. 3-9 ◽  
Author(s):  
Sandra Sülzenbrück
Keyword(s):  
Empirical Research ◽  
Visual Feedback ◽  
Closed Loop ◽  
Motor Planning ◽  
Visual Input ◽  
Closed Loop Control ◽  
Loop Control ◽  
Feedback Type ◽  
Effective Use ◽  
The Impact

For the effective use of modern tools, the inherent visuo-motor transformation needs to be mastered. The successful adjustment to and learning of these transformations crucially depends on practice conditions, particularly on the type of visual feedback during practice. Here, a review about empirical research exploring the influence of continuous and terminal visual feedback during practice on the mastery of visuo-motor transformations is provided. Two studies investigating the impact of the type of visual feedback on either direction-dependent visuo-motor gains or the complex visuo-motor transformation of a virtual two-sided lever are presented in more detail. The findings of these studies indicate that the continuous availability of visual feedback supports performance when closed-loop control is possible, but impairs performance when visual input is no longer available. Different approaches to explain these performance differences due to the type of visual feedback during practice are considered. For example, these differences could reflect a process of re-optimization of motor planning in a novel environment or represent effects of the specificity of practice. Furthermore, differences in the allocation of attention during movements with terminal and continuous visual feedback could account for the observed differences.

Sign in / Sign up

Export Citation Format

Share Document