Efficient integration of inertial observations into visual SLAM without initialization

Human body orientation estimation from microinertial/magnetic sensor units is highly important for synthetic environments, robotics, and other human-computer interaction applications. In practice, the main challenge is how to deal with linear acceleration interference and magnetic disturbance which always cause significant attitude-estimation errors. In this paper, we present a novel quaternion-based Kalman filter with vector selection scheme for accurate human body orientation estimation using an inertial/magnetic sensor unit. In the proposed algorithm, the gyroscope measurement is used as an input to construct the linear process equation, and the accelerometer and magnetometer measurements are manipulated to establish the linear pseudomeasurement equation. A linear Kalman filter is then deployed to estimate the body orientation. In the Kalman filter framework, a vector selection scheme is designed to protect the algorithm against undesirable conditions such as temporary intensive movement and magnetic disturbance and enable it to acquire more accurate orientation estimation. The experimental results have shown that the proposed algorithm can provide accurate attitude estimations with regard to the ground truth.

Accurate orientation estimation is a crucial component of 3D molecular structure reconstruction, both in single-particle cryo-electron microscopy (cryo-EM) and in the increasingly popular field of cryo-electron tomography (cryo-ET). The dominant approach, which involves searching for the orientation that maximizes cross-correlation relative to given templates, is sub-optimal, particularly under low signal-to-noise conditions. In this work, we propose a Bayesian framework for more accurate and flexible orientation estimation, with the minimum mean square error (MMSE) estimator serving as a key example. Through simulations, we demonstrate that the MMSE estimator consistently outperforms the cross-correlation-based method, especially in challenging low signal-to-noise scenarios, and we provide a theoretical framework that supports these improvements.When incorporated into iterative refinement algorithms in the 3D reconstruction pipeline, the MMSE estimator markedly improves reconstruction accuracy, reduces model bias, and enhances robustness to the “Einstein from Noise” artifact. Crucially, we demonstrate that orientation estimation accuracy has a decisive effect on downstream structural heterogeneity analysis. In particular, integrating the MMSE-based pose estimator into frameworks for continuous heterogeneity recovery yields accuracy improvements approaching those obtained with ground-truth poses, establishing MMSE-based pose estimation as a key enabler of high-fidelity conformational landscape reconstruction. These findings indicate that the proposed Bayesian framework could substantially advance cryo-EM and cryo-ET by enhancing the accuracy, robustness, and reliability of 3D molecular structure reconstruction, thereby facilitating deeper insights into complex biological systems.

PurposeThis paper aims to develop an adaptive unscented Kalman filter (AUKF) formulation for orientation estimation of aircraft and UAV utilizing low‐cost attitude and heading reference systems (AHRS).Design/methodology/approachA recursive least‐square algorithm with exponential age weighting in time is utilized for estimation of the unknown inputs. The proposed AUKF tunes its measurement covariance to yield optimal performance. Owing to nonlinear nature of the dynamic model as well as the measurement equations, an unscented Kalman filter (UKF) is chosen against the extended Kalman filter, due to its better performance characteristics. The unscented transformation of the UKF is shown to equivalently capture the effect of nonlinearities up to second order without the need for explicit calculations of the Jacobians.FindingsIn most conventional AHRS filters, severe problems can occur once the system suddenly experiences additional acceleration, resulting in erroneous orientation angles. On the contrary in the high dynamic accelerative mode of the new proposed filter the errors would not suddenly increase, since the additional to cruise accelerations are being continuously estimated resulting in substantially more accurate orientation estimation. This feature causes the associated filter errors to gradually increase, in the event of continuous vehicle acceleration, up to a point of zero additional acceleration that subsequently causes a subsidence of the error back to zero.Practical implicationsThe proposed filtering methodology can be implemented for orientation estimation of aircraft and UAV that are equipped with low‐cost AHRSs.Originality/valueTraditional AHRS algorithms utilize the accelerometers output for the computation of roll and pitch angles and magnetometer output for the heading angle. Moreover, these angles are also calculated from the gyroscopes output as well, but with errors that increase with time. Of course for some applications of AHRS system, orientation errors can be damped out with a proportional‐integral controller. In such a case, the filter cut‐off frequency is usually selected experimentally. But, for high accelerating vehicles utilizing AHRS, the system errors can become very large. A possible remedy to this problem could be to use more advanced nonlinear filter algorithms such as the one proposed.

Falling from a height, usually from a building, occurs ordinarily in suicide, in some accidents, and sometimes as an act of homicide. The point of trajectory, the horizontal distance and the impact point are closely related to the initial velocity, angle and height. This study examines the falling pattern in order to determine the mental status of the jumper as well as the manner of death. Initial velocity is found using horizontal movement and height. A serial study of athletes performing both the running jump (long jump) and standing jump (swimmer's start jump) via biomechanical methods is described. The initial velocity of the running jump and standing jump in normal athletics is 9.15 and 2.70 m/s with initial jumping angles of 21 and 38 deg, respectively. The maximal horizontal velocity of 9.15 m/s is closely related to maximal strength of initial velocity, angle of engaged force, and height. Theoretical estimation of the initial velocity between 2.70 and 9.15 m/s is correlative with the unique initial velocity and running jump to fall from a height that is closely related to the voluntary and attempted jump. Hence, the jumping victim with an initial velocity higher than 2.70 m/s implies suicide. These results indicate that horizontal distance and height are legitimate measures to use in speculating on the falling pattern and the manner of death. A unique case of suicide involving a run and jump with initial velocity greater than 2.70 m/s is illustrated.

Object-Oriented Environment for Assessing Tracking Algorithms

We propose a resource allocation algorithm combining transmit power and signal effective time-width joint optimization to improve the multi-target tracking performance and to achieve better accuracy of velocity estimation under the condition of limited system resources for MIMO radar. An optimization mode is established with two optimal parameters: the transmit power and the effective time of the signal to minimize the CRLB and solve this nonconvex optimization model by the sequential parameter convex estimation (SPCA) algorithm. Simulation results show that the proposed algorithm can significantly improve the estimation accuracy of target velocity and the effect on the accuracy of target velocity estimation of signal effective time-width is more than that of the transmitted power.

An Observer Design for Visual Simultaneous Localisation and Mapping with Output Equivariance

This study addresses challenges in sensor fusion for accurate and robust joint orientation estimation in human movement analysis using wearable inertial measurement units (IMUs). A magnetometer-free refined Kalman filter (KF) approach is presented and validated to address various indoor environmental constraints and challenges posed by human movement. These include variability in motion and dynamics, as well as magnetic disturbances caused by ferromagnetic materials or electronic interferences. Our proposed approach utilizes a Kalman-filter-based framework that analyzes the accelerometer’s alignment with the Earth’s frame to estimate orientation and correct gyroscope readings, eliminating reliance on magnetometer inputs. The algorithm was tested on both controlled robotic movements and real-world upper-limb-motion-monitoring scenarios. First, a comparative analysis was conducted on the double-stage Kalman filter (DSKF) and complementary filter using the collected robot motion encoder data. The results demonstrated superior performance in orientation estimation, particularly in yaw measurements, where the proposed method significantly improved accuracy. It achieved a lower root mean square error (RMSE = ) and mean absolute error (MAE = ), outperforming both the DSKF and complementary filter approaches. Additionally, the study’s findings were validated against a standard motion capture system, revealing error metrics within generally acceptable ranges ( of the joint range of motion [ROM]) and strong correlation coefficients (). However, some deviations were observed during complex motion cycle intervals, highlighting opportunities for further refinement. These findings suggest that the proposed approach presents a promising alternative for human joint orientation estimation in industrial settings with magnetic distortions.

Pedestrian Dead Reckoning (PDR) uses inertial measurement units (IMUs) and combines velocity and orientation estimates to determine a position. The estimation of the velocity is still challenging, as the integration of noisy acceleration and angular speed signals over a long period of time causes large drifts. Classic approaches to estimate the velocity optimize for specific applications, sensor positions, and types of movement and require extensive parameter tuning. Our novel hybrid filter combines a convolutional neural network (CNN) and a bidirectional recurrent neural network (BLSTM) (that extract spatial features from the sensor signals and track their temporal relationships) with a linear Kalman filter (LKF) that improves the velocity estimates. Our experiments show the robustness against different movement states and changes in orientation, even in highly dynamic situations. We compare the new architecture with conventional, machine, and deep learning methods and show that from a single non-calibrated IMU, our novel architecture outperforms the state-of-the-art in terms of velocity (≤0.16 m/s) and traveled distance (≤3 m/km). It also generalizes well to different and varying movement speeds and provides accurate and precise velocity estimates.

Recently, Simultaneous Localization and Mapping (SLAM) has achieved good performance in a certain environment. However, in a complex indoor environment, the tracking of the position of the mobile robot in Visual SLAM may get lost in some scenes with hardly any features. This paper presents a mobile robot localization method using an asynchronous data fusion formed by a wheel odometer and a visual odometer in the Extended Kalman Filter (EKF) based on the Semantic SLAM system. The first approach uses Semantic SLAM based on ORB-SLAM2 with YOLOv3, to get rid of the dynamic object while localizing and mapping, and build a global point cloud map where keyframes information is stored. Then, wheel odometer information and visual odometer information acquire after using a global point cloud map. Finally, the asynchronous data fusion for mobile robot localization, composed of visual odometer and wheel odometer, is realized at a frequency of about 50Hz in EKF. The experimental results on the real robot validate the method in accuracy and robustness by comparing it with ground truth from the motion capture measurement system in indoor environment.

This paper presents a vision-tracking system for mobile robots, which travel in a 3-dimentional environment. The developed system controls pan and tilt actuators attached to a camera so that a target is always directly in the line of sight of the camera. This is achieved by using data from robot wheel encoders, a 3-axis accelerometer, a 3-axis gyroscope, pan and tilt motor encoders, and camera. The developed system is a multi-rate sampled data system, where the sampling rate of the camera is different with that of the other sensors. For the accurate estimation of the robot velocity, the developed system detects the slip of robot wheels, by comparing the data from the encoders and the accelerometer. The developed system estimates the target position by using an extended Kalman filter. The experiments are performed to show the tracking performance of the developed system in several motion scenarios, including climbing slopes and slip cases.KeywordsVision tracking systemSensor data fusionKalman filterSlip detection

Rollover crashes are known as the most dangerous type of accidents throughout the world. They are associated with multi-directional velocities and accelerations which creates a complex dynamic behavior of the vehicle. Several experimental and numerical methods have been used to gain a better understanding of kinematics of the vehicle and occupants during the rollover crashes. Due to the complex nature of the rollover, any change in the initial conditions may significantly influence the rollover outcomes. The main goal of this study is to assess the effects of initial conditions on dynamic responses of the bus using a modified dolly rollover test procedure. Since the experimental rollover test is very expensive and to decrease the computational costs, the numerical model of cutaway bus was developed using lumped mass-spring-damper in PC-Crash software. First, the model was validated using the experimental data. Then, a series of simulations have been conducted with considering various initial conditions (inputs) such as initial velocity, friction of rollover surface, height of bus’s CG, and initial roll angle. The range of initial variables were selected based on Latin Hypercube Sampling (LHS) with uniform distribution. The simulation results were used to build the surrogate model using Kriging model for each rollover outcomes (outputs) including number of quarter turns, roll distance, deceleration rate, and maximum impact force. The sensitivity of the model to 400 set of input data was computed in MATLAB. The results of the sensitivity analysis indicate that the number of quarter turns and roll distance was highly affected by the initial velocity of the bus. Furthermore, the deceleration rate controlled mainly by friction and initial velocity and a negative correlation with the initial roll rate. No strong correlation between maximum impact force and any input parameters was observed. This led us to perform further research on exploring the effects of dynamic characteristics of the bus on maximum impact force.

Recognition of anatomical structures is an important step in model based medical image segmentation. It provides pose estimation of objects and information about "where" roughly the objects are in the image and distinguishing them from other object-like entities. In,¹ we presented a general method of model-based multi-object recognition to assist in segmentation (delineation) tasks. It exploits the pose relationship that can be encoded, via the concept of ball scale (b-scale), between the binary training objects and their associated grey images. The goal was to place the model, in a single shot, close to the right pose (position, orientation, and scale) in a given image so that the model boundaries fall in the close vicinity of object boundaries in the image. Unlike position and scale parameters, we observe that orientation parameters require more attention when estimating the pose of the model as even small differences in orientation parameters can lead to inappropriate recognition. Motivated from the non-Euclidean nature of the pose information, we propose in this paper the use of non-Euclidean metrics to estimate orientation of the anatomical structures for more accurate recognition and segmentation. We statistically analyze and evaluate the following metrics for orientation estimation: Euclidean, Log-Euclidean, Root-Euclidean, Procrustes Size-and-Shape, and mean Hermitian metrics. The results show that mean Hermitian and Cholesky decomposition metrics provide more accurate orientation estimates than other Euclidean and non-Euclidean metrics.

In dynamic environments, performance of visual SLAM techniques can be impaired by visual features taken from moving objects. One solution is to identify those objects so that their visual features can be removed for localization and mapping. This paper presents a simple and fast pipeline that uses deep neural networks, extended Kalman filters and visual SLAM to improve both localization and mapping in dynamic environments (around 14 fps on a GTX 1080). Results on the dynamic sequences from the TUM dataset using RTAB-Map as visual SLAM suggest that the approach achieves similar localization performance compared to other state-of-the-art methods, while also providing the position of the tracked dynamic objects, a 3D map free of those dynamic objects, better loop closure detection with the whole pipeline able to run on a robot moving at moderate speed.

Being able to measure 3-D flow velocity and volumetric flow rate effectively in the cardiovascular system is valuable but remains a significant challenge in both clinical practice and research. Currently, there has not been an effective and practical solution to the measurement of volume flow using ultrasound imaging systems due to challenges in existing 3-D imaging techniques and high system cost. In this study, a new technique for quantifying volumetric flow rate from the cross-sectional imaging plane of the blood vessel was developed by using speckle decorrelation (SDC), 2-D high-frame-rate imaging with a standard 1-D array transducer, microbubble contrast agents, and ultrasound imaging velocimetry (UIV). Through SDC analysis of microbubble signals acquired with a very high frame rate and by using UIV to estimate the two in-plane flow velocity components, the third and out-of-plane velocity component can be obtained over time and integrated to estimate volume flow. The proposed technique was evaluated on a wall-less flow phantom in both steady and pulsatile flow. UIV in the longitudinal direction was conducted as a reference. The influences of frame rate, mechanical index (MI), orientation of imaging plane, and compounding on velocity estimation were also studied. In addition, an in vivo trial on the abdominal aorta of a rabbit was conducted. The results show that the new system can estimate volume flow with an averaged error of 3.65% ± 2.37% at a flow rate of 360 mL/min and a peak velocity of 0.45 m/s, and an error of 5.03% ± 2.73% at a flow rate of 723 mL/min and a peak velocity of 0.8 m/s. The accuracy of the flow velocity and volumetric flow rate estimation directly depend on the imaging frame rate. With a frame rate of 6000 Hz, a velocity up to 0.8 m/s can be correctly estimated. A higher mechanical index (MI = 0.42) is shown to produce greater errors (up to 21.78±0.49%, compared to 3.65±2.37% at MI = 0.19). An in vivo trial, where velocities up to 1 m/s were correctly measured, demonstrated the potential of the technique in clinical applications.