Real-time Estimation Method Research Articles

Diverse Humanoid Robot Pose Estimation from Images Using Only Sparse Datasets

We present a novel dataset for humanoid robot pose estimation from images, addressing the critical need for accurate pose estimation to enhance human–robot interaction in extended reality (XR) applications. Despite the importance of this task, large-scale pose datasets for diverse humanoid robots remain scarce. To overcome this limitation, we collected sparse pose datasets for commercially available humanoid robots and augmented them through various synthetic data generation techniques, including AI-assisted image synthesis, foreground removal, and 3D character simulations. Our dataset is the first to provide full-body pose annotations for a wide range of humanoid robots exhibiting diverse motions, including side and back movements, in real-world scenarios. Furthermore, we introduce a new benchmark method for real-time full-body 2D keypoint estimation from a single image. Extensive experiments demonstrate that our extended dataset-based pose estimation approach achieves over 33.9% improvement in accuracy compared to using only sparse datasets. Additionally, our method demonstrates the real-time capability of 42 frames per second (FPS) and maintains full-body pose estimation consistency in side and back motions across 11 differently shaped humanoid robots, utilizing approximately 350 training images per robot.

A real-time estimation method of carbon emission for cement corporations based on load identification

Abstract Accurate carbon emission estimation is the basis for assessing the carbon emissions of various regions and industries and formulating reasonable and feasible emission reduction programs. This paper presents a generalized real-time corporate carbon footprint measurement method. The total carbon emissions are categorized into direct emissions and indirect emissions. For the former, a load identification method based on CNN-BLSTM is proposed to monitor the state of the devices, and the carbon emissions are directly calculated in combination with the carbon emission intensity of the devices; for the latter, accurate estimation of carbon emissions is achieved through the application of marginal carbon emission factor. By utilizing this analytical model, the estimated carbon emissions are more accurate than those of other methods, which can better grasp the carbon emissions of enterprises and promote energy saving and emission reduction.

Real-Time Analysis of Facial Expressions for Mood Estimation

This paper proposes a model-based method for real-time automatic mood estimation in video sequences. The approach is customized by learning the person’s specific facial parameters, which are transformed into facial Action Units (AUs). A model mapping for mood representation is used to describe moods in terms of the PAD space: Pleasure, Arousal, and Dominance. From the intersection of these dimensions, eight octants represent fundamental mood categories. In the experimental evaluation, a stimulus video randomly selected from a set prepared to elicit different moods was played to participants, while the participant’s facial expressions were recorded. From the experiment, Dominance is the dimension least impacted by facial expression, and this dimension could be eliminated from mood categorization. Then, four categories corresponding to the quadrants of the Pleasure–Arousal (PA) plane, “Exalted”, “Calm”, “Anxious” and “Bored”, were defined, with two more categories for the “Positive” and “Negative” signs of the Pleasure (P) dimension. Results showed a 73% of coincidence in the PA categorization and a 94% in the P dimension, demonstrating that facial expressions can be used to estimate moods, within these defined categories, and provide cues for assessing users’ subjective states in real-world applications.

U-HRNet+: a real-time precipitation estimation method based on the fusion of temporal and spatial domain features

ABSTRACT The precipitation estimation with high spatial and temporal resolution over Tibet is of great significance for monitoring the stability of the plateau ecosystem and early warning of natural disasters. In response to the problem of the unsatisfactory accuracy in traditional real-time precipitation inversion methods and quantitative accuracy estimation products in this region. This paper adopts deep-learning methods to explore the corresponding relationship between precipitation and multi-band observation data from geostationary satellite imagery. The U-High Resolution Network (U-HRNet) as backbone is fully exploited to generate multi-scale features and fused using attention gates to extract high-resolution spatial representations representing precipitation clouds, thereby achieving precipitation estimation of target clouds at different scales. In addition, in response to the characteristics of high precipitation intensity and drastic cloud cluster changes in summer convective weather within a short period of time, a multi-head attention mechanism is introduced to extract temporal information from time series inputs and fuse it with spatial features to improve the accuracy of real-time precipitation estimation. Compared with the operational precipitation products and baseline deep-learning models, the classification metrics such as Probability of detection (POD), False Alarm Ratio (FAR) and Critical success index (CSI) show that the proposed model (U-HRNet+) has better performance in test set and has high generalization ability for estimating short-term heavy rainfall over Tibet and its surrounding areas in summer. Therefore, the proposed model has the potential to serve as a more accurate and reliable satellite-based precipitation estimation product.

AM-HNN based real-time voltage estimation method of distribution network using joint PMU and AMI data

A large amount of distributed energy access increases the risk of voltage fluctuation and off-limit voltage. Accurate voltage estimation is more and more important to ensure the safety of distribution networks. With the introduction of phasor measurement unit (PMU), multi-source measuring data with different time scales and measuring precision exist in distribution networks. How to fully integrate multi-source measurement data to achieve fast and accurate state estimation of distribution networks is an important prerequisite for efficient operation decision making. In this paper, a distribution network voltage state estimation method based on AM-HNN model is proposed. The attentional mechanism (AM) is introduced to mine the influence of PMU measurement data on voltage state estimation, and the feature weight is modified adaptively. The highway neural network (HNN) was used to fit the mapping relationship between the injected power of grid nodes and the voltage of key nodes. Taking the distribution network data of different scenes as an example, the superiority of the proposed optimal measurement model is verified by comparing with various voltage estimation methods.

HDA-pose: a real-time 2D human pose estimation method based on modified YOLOv8

HDA-pose: a real-time 2D human pose estimation method based on modified YOLOv8

A real-time hydrogen consumption estimation method for fuel cell vehicles

ABSTRACT As an economic evaluation indicator for fuel cell vehicles (FCVs), real-time hydrogen consumption’s accuracy and immediateness are important. Real-time hydrogen consumption obtained through the fuel cell system (FC Power method) has strong immediateness but poor accuracy, while through the hydrogen mass (Mass method) has good accuracy but poor immediateness. Given the difficulty in obtaining sufficient training data on actual vehicles, this paper establishes a hydrogen storage system model whose input and output can be used to obtain data. Then, a new estimation method based on the prediction of the mass’s difference (PMD method) has been constructed based on the Neural basis expansion analysis with exogenous variables (NBEATSx) algorithm. After the NBEATSx algorithm is trained, the prediction accuracy of the PMD method is verified under urban and high-speed conditions. The result shows that the new method proposed in this article improves Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) by nearly 45% compared to the FC Power method, and the hydrogen consumption’s sampling point is advanced compared to the Mass method. The trained model can be input into Hydrogen Management System (HMS) which is installed on the hydrogen storage system and output data in real-time.

DTP: learning to estimate full-body pose in real-time from sparse VR sensor measurements

For virtual reality (VR) applications, estimating full-body pose in real-time is becoming increasingly popular. Previous works have reconstructed full-body motion in real time from an HTC VIVE headset and five VIVE Tracker measurements by solving the inverse kinematics (IK) problem. However, an IK solver may yield unnatural poses and shaky motion. This paper introduces Deep Tracker poser (DTP): a method for real-time full-body pose estimation in VR. This task is difficult due to the ambiguous mapping from the sparse measurements to full-body pose. The data obtained from VR sensors is calibrated, normalized and fed into the deep neural networks (DNN). To learn from sufficient data, we propose synthesizing a VR sensor dataset called AMASS-VR from the AMASS, a collection of various motion capture datasets. Furthermore, feet tracking loss is a common problem of VIVE Tracker. To improve the accuracy and robustness of DTP to the occlusion noise, we simulate the occlusion noise by Gaussian random noise. Then we synthesize an occlusion dataset AMASS-OCC and fine-tune DTP on that. We evaluate DTP by comparing with other popular methods in terms of the accuracy and computational cost. The results indicate that DTP outperforms others in terms of the positional error (1.04 cm) and rotational error (4.22 °). The quantitative and qualitative results show that DTP reconstructs accurate and natural full-body pose even under serious feet occlusion, which indicates the superiority of the DTP in modelling the mapping from sparse joint data to the full-body pose.

Real-time estimation of fuel injection rate and injection volume in high-pressure common rail systems

The injection rate and injection volume are essential parameters for the combustion process, which greatly affect the engine efficiency and emission performance. However, at the current time the injection information cannot be obtained in real time during the actual operation of engine. This paper presents a novel real-time estimation method for the injection rate and injection volume based on the measurable rail pressure. A comprehensive dynamic model is derived by considering both the effect of fuel injection and supply process on the pressure fluctuation. Then a linear time-varying state space model is constructed by choosing three appropriate state variables. On this basis, considering the measurement noise and model uncertainty, a Kalman filter-based estimation method of the injection information is investigated. And the noise covariance matrix Q is optimally designed to adapt to a wide range of operating conditions. The proposed estimation method is validated on the different conditions, and the results show the injection rate estimation errors are within 4.5 %, the injection volume estimation errors are within 4 %. And on the situation with overlap of injection and supply process, the injection rate and injection volume estimation errors are all within 4.5 %.

An enhanced real-time human pose estimation method based on modified YOLOv8 framework

The objective of human pose estimation (HPE) derived from deep learning aims to accurately estimate and predict the human body posture in images or videos via the utilization of deep neural networks. However, the accuracy of real-time HPE tasks is still to be improved due to factors such as partial occlusion of body parts and limited receptive field of the model. To alleviate the accuracy loss caused by these issues, this paper proposes a real-time HPE model called CCAM-Person\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\ extbf {CCAM-Person}}$$\\end{document} based on the YOLOv8 framework. Specifically, we have improved the backbone and neck of the YOLOv8x-pose real-time HPE model to alleviate the feature loss and receptive field constraints. Secondly, we introduce the context coordinate attention module (CCAM) to augment the model’s focus on salient features, reduce background noise interference, alleviate key point regression failure caused by limb occlusion, and improve the accuracy of pose estimation. Our approach attains competitive results on multiple metrics of two open-source datasets, MS COCO 2017 and CrowdPose. Compared with the baseline model YOLOv8x-pose, CCAM-Person improves the average precision by 2.8% and 3.5% on the two datasets, respectively.

Research on state-of-charge estimation of lithium-ion batteries based on an improved gas-liquid dynamics model

Improving the accuracy of state of charge (SOC) is beneficial for the safety and service life of batteries. The battery analysis model has become one of the research hotspots due to its clear physical significance and visibility of parameters. Based on analytical modeling methods, an improved gas-liquid dynamics (GLD) model of the lithium-ion battery (LIB) is proposed, which excels in simulating lithium-ion diffusion behavior and terminal voltage rebound characteristics under the open circuit state of LIBs. The parameter mapping relationships between the LIB and GLD model are improved and estimation equations of SOC, terminal voltage, and open circuit voltage are deduced according to GLD state equations. Based on these estimation equations and dual extended Kalman filter, an efficient real-time SOC estimation method for LIBs is developed. The performance of SOC estimation method is verified by LIBs data sampled by onboard BMS under five working conditions. Results show that the proposed method has strong robustness of the initial error of SOC = ±20 % which is eliminated within 15 s, and has a reliable performance of SOC estimation against the current and terminal voltage noises.

Infrastructure 2D Camera-based Real-time Vehicle-centered Estimation Method for Cooperative Driving Support

Infrastructure 2D Camera-based Real-time Vehicle-centered Estimation Method for Cooperative Driving Support

Multilayer Perceptron-Based Error Compensation for Automatic On-the-Fly Camera Orientation Estimation Using a Single Vanishing Point from Road Lane

This study introduces a multilayer perceptron (MLP) error compensation method for real-time camera orientation estimation, leveraging a single vanishing point and road lane lines within a steady-state framework. The research emphasizes cameras with a roll angle of 0°, predominant in autonomous vehicle contexts. The methodology estimates pitch and yaw angles using a single image and integrates two Kalman filter models with inputs from image points (u, v) and derived angles (pitch, yaw). Performance metrics, including avgE, minE, maxE, ssE, and Stdev, were utilized, testing the system in both simulator and real-vehicle environments. The outcomes indicate that our method notably enhances the accuracy of camera orientation estimations, consistently outpacing competing techniques across varied scenarios. This potency of the method is evident in its adaptability and precision, holding promise for advanced vehicle systems and real-world applications.

Real-time estimation of barycenter position for fuel cell distributed drive electric tractor based on derived UKF

ABSTRACT The FCDET is considered as a promising and advanced agricultural machinery due to its convenience and controllability. To enhance the control performance of the FCDET unit dynamic system, understanding the accurate position of the center of gravity (CoG) of the tractor unit in longitudinal motion and the operating environment is essential. In this paper, we proposed a real-time estimation method of CoG position of longitudinal motion for FCDET plowing and transportation conditions. CoG height and longitudinal axis distance distribution coefficients are used as state estimation variables, and the data obtained from displacement and velocity sensors are used as state observation variables, establishing a discrete system state space model and proving the observability. The proposed Derived UKF algorithm introduces the Sage-Husa adaptive noise equation within the Square-Root UKF framework to address the non-positive definite matrix problem. The research shows that the precise CoG longitudinal position effectively reflects changes in operational pavement load. The Derived UKF algorithm exhibits high state estimation accuracy in plowing and transportation conditions, with mean absolute errors of 14.2% and 5.0% of the state amplitude, respectively. This study supports the practical application of real-time state parameter estimation algorithms in controlling advanced agricultural machinery distributed drive systems.

Real-time prediction of material removal rate for advanced process control of chemical mechanical polishing

Polishing torque holds significance in monitoring the chemical mechanical polishing (CMP) process due to its close correlation with material removal. This study introduces a new model-based technique for estimating the material removal rate (MRR) using in-process data from CMP machines. The proposed method employs either the sequential least squares method or Kalman filter for real-time state estimation. Real-time estimation of MRR enables material removal control without relying on conventional endpoint detection (EDP) technologies. The accuracy of the proposed approach is validated through oxide CMP experiments, demonstrating precise estimation of the center-slowed MRR profile towards the end of the pad life.

Real-Time Optical Flow Estimation Method Based on Cross-Stage Network

In this paper, a real-time optical flow estimation method based on a cross-stage network is proposed. The proposed model is designed with a network structure with encoders and decoders. The proposed method combines cross-stage network technology with the network structure of FlowNet2 and RAFT to achieve improved parameter number and estimation performance. For real-time optical flow estimation, it is important to maintain performance while reducing the number of parameters in the network. In the proposed method, structural convergence is performed to increase performance while reducing the number of parameters by applying the cross-stage network structure. The proposed model is designed to solve the bottlenecks in model accuracy and complexity by separating feature extraction and flow estimation processes. Flying Chairs, Flying Things 3D, and KITTI datasets were used to evaluate the performance of the proposed model, and the experimental results show superior performance compared to previous traditional methods.

Real-time clock estimation using system time offset maintenance aiming at multi-GNSS time interoperability

Time interoperability is one of the important aspects of Global Navigation Satellite System (GNSS) compatibility and interoperability, which will greatly benefit the accuracy and availability of the positioning, navigation and timing (PNT) solution. In this paper, we proposed a real-time satellite clock estimation method using system time offset (STO) maintenance to help users realize GNSS time interoperability. First, receiver inter-system bias (ISB) was investigated and modeled for STO maintenance. Then, we presented clock estimation assessment results, which demonstrate that the new method can make the satellite clock product maintain a stable STO. Finally, precise point positioning (PPP) was used for further validation. The results show that in the condition of a 45° cutoff elevation angle, fixing the ISB can significantly reduce the convergence time, increase the availability by 21.5%, and improve the average positioning accuracy in horizontal and vertical by 37.7% and 44.4%, respectively.

A Real-Time Permittivity Estimation Method for Stepped-Frequency Ground-Penetrating Radar by Full-Waveform Inversion

Ground-penetrating radar (GPR) has been widely used in estimating the permittivity of mediums. The radar echo amplitude method is an important method used by GPR in this estimation, the basic step of which is to deduce the magnitude of the permittivity according to the relationship between the reflection coefficient and the permittivity. Based on the basic principle of the radar echo amplitude method, this paper proposes a full-wave inversion real-time permittivity estimation method that can be used for stepped-frequency GPR (SFGPR), which offers high efficiency, accuracy, and generalization ability. The characteristics of this method are mainly reflected in the following four aspects: Using the SFGPR system and introducing a layered media detection model, we can complete waveform compensation optimization with high precision. The distance between the antenna and the surface of the reflective medium is extracted from the time domain waveform without manual measurement, avoiding human measurement errors. The inversion of the total reflection waveform at the required height works under the principle of an electromagnetic field, eliminating the need for repeated metal plate calibration experiments and improving work efficiency and waveform accuracy. In a continuous measurement line, the total reflection waveform inversion on each measurement point can be efficiently completed, and the change of permittivity on the measurement line can be obtained. To evaluate the feasibility of the proposed method, we performed experiments on a wall of known thickness, and the permittivity estimation was basically consistent with that of the dielectric probe, physical model calculation, and wall penetration. We also successfully applied this method to the dielectric property analysis of adobe samples. The results indicate that the proposed method can help grasp the condition of a measured medium, which can ensure the accuracy of detection and improve subsequent data processing efficiency.

Real-time tsunami forecasting system with nonlinear effects using Green’s functions: application to near-shore tsunami behavior in complex bay topography

ABSTRACT In this study, a real-time tsunami estimation method was developed using the linear superposition of Green’s functions, which account for nonlinear tsunami effects around coastal areas. To this end, first, the nonlinear effects associated with the inundation and bottom friction and advection terms in a nonlinear long-wave model were investigated, focusing on large tsunamis around Urado Bay in Kochi Prefecture, Japan, which is at risk of future damage from tsunamis produced by the Nankai Trough earthquake. The results indicate that the effects of advection play a more significant role than those of the inundation and bottom friction in terms of accurately estimating near-shore tsunami propagation. Then, based on these results, Green’s functions were modified to account for the linearly approximated effects of the advection terms. By applying the developed method to tsunamis around Urado Bay, it was demonstrated that accurate estimations of the water surface level and flow velocity changes could be quickly obtained at any location around the bay. The developed method can be applied to real-time tsunami inundation forecasting and off-shore evacuation simulations of vessels to support optimal evacuation decisions.

A real-time spatial deformation estimation method based on spatial curvature decomposition and interpolation

Structural deformation entails safety risks, and at times, direct measurement can be challenging, especially when reference points are unavailable. Therefore, the development of a practical and precise deformation estimation method is crucial. This paper introduces a strain-based method with spatial curvature decomposition and interpolation for estimating structural deformation. The proposed method was tested through numerical simulations of various static and dynamic deformation cases for different structures, including a railway track, a rigid hanger of a bridge under low-frequency dynamic loads, and a cantilever beam under high-frequency dynamic loads. The feasibility of the method was verified, and it was further validated by a field test on a railway track. Compared to existing strain-based methods, it offers more flexible sensor placement, requires fewer parameters, and can estimate dynamic, static, planar, and spatial deformations. This strain-based method shows promise for real-time structural deformation monitoring, enhancing safety assessments.