Lab Motion Research Articles

A Deep Convolutional Autoencoder for Automatic Motion Artifact Removal in Electrodermal Activity.

This study aimed to develop a robust and data driven automatic motion artifacts (MA) removal technique from electrodermal activity (EDA) signal. we proposed a deep convolutional autoencoder (DCAE) approach for automatic MA removal in EDA signals. Our model was trained using several publicly available datasets that were collected using a wide variety of stimuli to cause EDA reactions; the sample size was large ([Formula: see text]). We trained and validated our DCAE network using both Gaussian white noise (GWN) and realistic MA data records collected using a novel circuitry in our lab. We further evaluated and compared the performance of our DCAE model with the existing methods on two independent and unseen datasets called Chon lab motion artifact dataset II (CMAD II) and central nervous system oxygen toxicity dataset (CNS-OT). Our DCAE model showed significantly higher signal-to-noise-power-ratio improvement ( SNRimp) and lower mean squared error ( MSE) when compared with that of the three previous methods (averaged [Formula: see text], and MSE = 0.028 on the MA-corrupted data). Moreover, the reconstructed EDAs from the CMAD II dataset had a mean correlation value of 0.78 (statistically significantly higher when compared with other methods) with the reference clean data from the motionless hand, whereas the raw MA-corrupted data had a correlation value of only 0.68. The results presented in the paper indicates that our DCAE can remove MAs with higher intensity where the existing methods fails. Proposed DCAE model can be used to recover a significant amount of otherwise discarded EDA data.

Design and Stability Analysis of High-Speed Unmanned Aerial Vehicle Electric Anti-Skid Braking System

A new combined electric anti-skid braking system of a high-speed unmanned aerial vehicle is designed to improve the stability and to reduce the system sensor noise. An electric anti-skid braking mechanism model considering the sensor noise effect is built in MATLAB/Simulink. The stability of a slip ratio braking system and a deceleration braking system is analyzed using the Routh criterion and Lyapunov stability method. Then the UAV ground taxiing dynamic model is built in LMS Virtual. Lab Motion. The braking performance and dynamic responses are studied under the control of the designed braking system using a co-simulation method. Conclusions show that the designed braking control system can solve the sensor noise problem effectively and the stability and robustness are both ensured under this control law.

Distribution of driving force beneath wheeled vehicle with varying center of gravity

Driving force analysis is performed on the no-spin differential and full-time all-wheel-drive vehicle; this thesis takes an automatic loading mixing vehicle as an example to introduce the compositions and working principle of the driving system. Based on the tire-ground mechanics, the model of the dynamics and the kinematics is established under the walking straight and steering conditions. According to the theoretical model, the influence of the vehicle’s gravity center on the moving system is analyzed. Co-simulation based on LMS Imagine Lab AMESim and LMS Virtual Lab Motion is performed to build the hydraulic driving system and the multi-body dynamics system models. Based on the tire-ground load environment simulation model built by 1D + 3D, various positions of the gravity center of the model are set to compare with the theoretical analysis. Various weight blocks are also added to change the location of the gravity center in the practical experiment. The conclusions that different gravity center positions lead to the change of the driving torque distribution are proved by the simulation results and experimental data.

Multi-body Dynamics Simulation and Analysis of Inertia Release Mechanism

For multiple work sections and complicated work conditions, Inertia Release Mechanism contains complicated contact surfaces of parts, it is hard to design and analyze with traditional methods. In this paper, the detailed structure model of Inertia Release Mechanism was designed with UG, and its virtual prototype model was established with LMS Virtual. Lab Motion. Based on the model, system kinematics and dynamics were simulated and analyzed. The simulation results revealed that with virtual prototype technology, the design level and efficiency of Inertia Release Mechanism can be enhanced. The model can also be used for structure optimum design.

Dynamical Analysis of an Undulatory Wheeled Locomotor: a Trident Steering Walker

This paper presents a novel methodology of dynamical analysis of an undulatory wheeled locomotor: a trident steering walker based on Lagrangian mechanics. The trident steering walker moves by undulatory locomotion in which it transforms its periodic changes in shape into its displacement and it is a mechanical system subject to nonholonomic constraints in which all wheels roll without slipping on a horizontal plane. The dynamical equations of the trident steering walker consist of the Euler-Lagrange equations and the time derivatives of the equations of the nonholonomic constraints. The nonholonomic constraints are especially incorporated through the Lagrangian undetermined multipliers. I demonstrate the effectiveness of a dynamical control law which is derived from a path-following feedback kinematical control law previously designed by the author by performing a path-following motion in which the trident steering walker follows a third-order Bezier curve path. The dynamical equations are numerically integrated and the simulated state variables of Lagrangian mechanics exactly correspond with those of a commercial multibody dynamics simulator, LMS Virtual. Lab Motion. Therefore, the validity of the dynamical analysis of Lagrangian mechanics is verified by the comparison with the commercial multibody dynamics simulator.

Evolutionary optimization of video event classification

Detection and classification of semantic events in video documents have been major tasks in the domain of automatic video analysis. The multimodality of video data brings about challenging issues and the effectiveness of its automatic semantic processing has been hampered by the video semantic gap, that is, the gap between low-level visual and spatio-temporal features used as a digital representation of video documents and their semantic interpretation. Traditionally, these low-level features are concatenated in high-dimensional spaces and classified as high-level semantic events through computational learning methods. Support vector machines (SVMs), widely used in the literature, have often shown to outperform other popular learning techniques for this task. However, concatenation of several features with different intrinsic properties and wide dynamical ranges may result in curse of dimensionality and redundancy issues. Among the factors impacting the effectiveness of classification are (i) feature subset selection, (ii) tuning of classifiers’ parameters and (iii) selection of proper training instances. In this paper, we address these factors and propose a technique (denoted as GAoptSVM) for an optimal SVM-based video event classification through the use of an evolutionary optimization technique. Extensive experiments on the 50 h video data set of TRECVid 2008 event detection task and large quantities of video data collected from Youtube and CMU Graphics Lab Motion Capture Database demonstrate that our approach outperforms the traditional SVM and effectively classify video events with noticeable accuracy.

Exploring new fields of virtual load prediction by accurate tire simulation for large deformations and flexible rim support

AbstractDuring the past ten years, there has been a significant trend in automotive design using low aspect ratio tires and increasingly run‐flat tires as well. In recent publications, the influence of those tire types on the dynamic loads – transferred from the road through the wheel into the car – have been examined pretty extensively, including comparative wheel force transducer measurements as well as computational results. It can be shown that the loads to the vehicle tend to increase when using low aspect ratio tires and particularly when using run‐flat tires. These tires provide higher stiffnesses while simultaneously introducing larger nonlinearities in the sidewall behavior [1–3]. Depending on manufacturer and the combination of vehicle size and wheel properties, these deformations can be so large that the tire belt and/or sidewall have contact with the rim crown (protected by the tire sidewall). The full vehicle simulation on virtual proving grounds is well established and important for the vehicle product development process. One of the most important subsystems in the virtual load assessment process, using full vehicle simulation is the tire model. The precision of that is essential for the overall accuracy of the virtual method. So the tendencies described above strongly require adaptations and improvements in the field of tire modeling. In 2007, Fraunhofer LBF together with Honda R&D started to examine the influences of low aspect ratio and run flat tires for the accuracy of full vehicle simulation results for durability relevant scenarios [3–5]. Those activities were the starting point for a four years joint activity to extend the usability of the virtual load prediction method by full vehicle simulation to application for which strong nonlinearities in the tire (large very transient deformation), but also in the vehicle model itself occur. As a part of that joint development, this paper summarizes the activities of Fraunhofer LBF to develop a dedicated tire model, which can accurately handle very large deformations of the tire up to misuse‐like applications. The model is based on the LBF tire model CDTire. In the first chapter several nonlinear extensions of the belt and sidewall model will be described which have been implemented to capture the large deformation behavior. These model extensions are also taking into account the belt‐to‐sidewall and sidewall‐to‐rim contact. To validate and to parameterize these model extensions, Fraunhofer LBF built a dedicated flat track test rig, which can be used to realize “roll‐over‐cleat” experiments using huge obstacles, so that belt‐to‐sidewall and sidewall‐to‐rim contact can be forced. This test rig will be described in chapter 2. The third chapter is dealing with the interface of the new tire model to a flexible rim. While the load transfer from road via tire into the vehicle is relatively easy to detect, for example by using wheel force transducers, the local forces acting on the rim flanges as well as on the wheel well (when e. g. passing a curb) are much more difficult to detect (in measurement as well as in simulation). LBF developed a method to detect local tire‐to‐rim interface forces and manage flexible rim simulation in Multi‐Body‐Simulation (LMS Virtual. Lab Motion – [6]). One key issue of the overall method is the capability of the tire model to predict local rim forces on the rim flanges in a suitable way. The second key issue is to combine the tire with a model of a flexible rim (which is embedded in a full vehicle MBS model). This method can be used to perform virtual load prediction of local, transient rim forces, which are the basis for CAE based fatigue life prediction of wheels applying typical durability test track and abuse load events.

A Landing Gear Drop Dynamic Simulation Based on the LMS. Virtual. Lab

In order to evaluate the dynamic behavior of the buffer of the Seagull 300 aircraft’s main landing gear, a drop model is built to simulate the drop dynamics using the software of LMS Virtual Lab Motion. The fluid damping force of the buffer, the air spring force of the buffer, the tire force of the landing gear and the weight of the fuselage are considered in the model. The simulation results are compared with the results of the Seagull 300 landing gears drop test, which proves the accuracy of the simulation model. Then the buffer performance and its influence factors are computationally discussed. This method gives a new way to study and improve the performance of the buffer system of an aircraft.

3D Human Pose Estimation from a Monocular Image Using Model Fitting in Eigenspaces

Generally, there are two approaches for solving the problem of human pose estimation from monocular images. One is the learning-based approach, and the other is the model-based approach. The former method can estimate the poses rapidly but has the disadvantage of low estimation accuracy. While the latter method is able to accurately estimate the poses, its computational cost is high. In this paper, we propose a method to integrate the learning-based and model-based approaches to improve the estimation precision. In the learning-based approach, we use regression analysis to model the mapping from visual observations to human poses. In the model-based approach, a particle filter is employed on the results of regression analysis. To solve the curse of the dimensionality problem, the eigenspace of each motion is learned using Principal Component Analysis (PCA). Finally, the proposed method was estimated using the CMU Graphics Lab Motion Capture Database. The RMS error of human joint angles was 6.2 degrees using our method, an improvement of up to 0.9 degrees compared to the method without eigenspaces.

Weaving Fabric at Unequalled Industrial Speed

Picanol, the leading weaving machine manufacturer, cooperated with LMS Engineering Services to realize higher productivity and enhanced noise and vibration standards on its next-generation rapier weaving machines. After providing support to Picanol's successful GamMax weaving machines, LMS optimized the new rapier driver mechanism for Picanol's upcoming rapier loom. Dynamic evaluations in LMS Virtual. Lab Motion helped Picanol and LMS engineers to reduce internal peak forces, avoid bearing wear-out, lower weight of oscillating parts, and extend the fatigue life of key parts. This enabled Picanol to significantly speed up the motion of the rapier as it travels through warp yarns, translating into 15% higher weaving productivity.