Modulating Dynamic Models for Lip Motion Generation

Linear vibration of thin-walled beams of equal angle-section

The nonlinear damping effect on response of coupled three degree-of-freedom autoparametric vibration system with elastic pendulum attached to the main mass is investigated numerically. It was assumed that the main body is suspended by an element characterized by non-linear elasticity and non-linear damping force and is excited harmonically in the vertical direction. The elastic pendulum characterized also by -linear elasticity and non-linear damping. Solutions for the system response are presented for specific values of the uncoupled normal frequency ratios and the energy transfer between modes of vibrations is observed. Curves of internal resonances for free vibrations and external resonances for exciting force are shown. In this type system one mode of vibration may excite or damp another one, and except different kinds of periodic vibration there may also appear chaotic vibration. Various techniques, including chaos techniques such as bifurcation diagrams and: time histories, phase plane portraits, power spectral densities, Poincare` maps and exponents of Lyapunov, are used in the identification of the responses. These bifurcation diagrams show many sudden qualitative changes, that is, many bifurcations in the chaotic attractor as well as in the periodic orbits. The results show that the system can exhibit various types of motion, from periodic to quasi-periodic to chaotic, and is sensitive to small changes of the system parameters.

Robotic motion generation methods using machine learning have been studied. Bilateral control-based imitation learning can imitate human motions using force information. Using this method, variable speed motion generation that con-siders physical phenomena such as the inertia and friction can be achieved. However, the previous study focused on a simple reciprocating motion. To learn the complex relationship between the force and speed more accurately, it is necessary to learn multiple actions using many joints. In this paper, we propose a variable speed motion generation method for multiple motions. We considered four types of neural network models for the motion generation and determined the best model for multiple motions at variable speeds. Subsequently, we used the best model to evaluate the reproducibility of the task completion time for the input completion time command. The results revealed that the proposed method could change the task completion time according to the specified completion time command in multiple motions.

A generic kinematic model is presented for a unified hydrodynamics study covering the various types of motion found in real as well as hypothetical fish-like undulation. Undulating motion is presented here as a generic motion, considering chordwise forced flexibility—modelled by wavelength of undulation λ*—of a rigid NACA0012 hydrofoil in a free-stream flow. Using a level-set immersed boundary method-based in-house code, a non-dimensional study is presented for various wavelengths, λ* (0.8–8.0), and frequency of undulation, St (0.2–0.7), at a constant maximum amplitude of undulation of 0.1 and Reynolds number of 5000. A unified cause-and-effect-based analysis is presented with the help of flow patterns and propulsive performance parameters. Pressure contour demonstrates how the travelling wave distributes momentum in the streamwise direction and reduces the lateral force coefficient. Vorticity contours elucidate the mechanism of formation of reverse von Kármán vortex street and the secondary vortices. A correlation is proposed for thrust coefficient as a function of λ* and St. Good qualitative agreement is observed between the simulated results at smaller λ*-based undulating and larger λ*-based pitching foil and the published results for the anguilliform and thunniform fishes, respectively. The agreement with the real fishes is presented for the relative magnitude of thrust coefficient, propulsive efficiency, dynamic stabilization, and signal of the prey fish for the predator fish. Similar to propulsive performance of real fishes, larger (smaller) flexibility-based undulation (pitching) results in larger propulsive efficiency (thrust generation)—which can be used for the design of fish-like biomimetic propulsion system.

Human motion generation has become an important research direction in computer vision and human motion modelling. Current motion generation methods typically rely on static or random masking during training, which fail to adequately capture dynamic variations in joint movement amplitude and temporal characteristics, resulting in suboptimal accuracy in generated motions. To address this, we propose a dynamic masking strategy (DMS) based on motion amplitude, which dynamically adjusts the mask probability distribution by incorporating both motion amplitude and temporal features. By calculating the motion amplitude of each joint and adapting the mask timing, the model is directed to focus on key movements during training, enhancing the quality of motion generation. Experimental results demonstrate that DMS outperforms traditional methods across multiple evaluation metrics, achieving a 15.3% reduction in FID, a 9.0% reduction in trajectory error, and a 6.3% reduction in location error, thereby validating the effectiveness and sophistication of the proposed method.

Modelling how proteins bind to one another is a challenging task, in part due to the conformational changes that can occur upon complex formation. We here describe the various types of motion that occur upon protein binding and discuss their treatment in computational protein-protein docking methods to predict the structure of the bound complex. Considered are five different categories of structural change, that cover timescales from picoseconds to milliseconds and amplitudes extending to tens of angstroms. First small-scale motion, which includes bond stretching, bond angle bending and dihedral rotation is addressed. This is followed by larger motions of the protein main-chain, the loops and entire protein domains. Finally, we consider the class of intrinsically disordered proteins including protein segments that refold upon binding. For each category, the capabilities and limitations of current docking procedures are discussed by means of an illustrative example.

Many livings such as fish and birds form groups which we consider as collective systems. The collective motion of the group shows various dynamics and patterns, and many model equations are proposed to explain such phenomenon. In our previous paper, we proposed simple deterministic model which shows various types of group behaviors. In this paper, we modified our model from the viewpoint of real robotic system. We assumed that the space resolution of each robot's sensors is low and that the sensors detect the nearest robots. We examined their performance by computer simulation and confirmed the robots with modified model also shows various types of motion. We also applied this system to foraging task and showed its performance.

This paper presents the development of the animation of robots in virtual reality environments, whose mechanisms can be coupled -the movement relies on mechanical principles-; and uncoupled mechanisms, i.e., the degrees of freedom are controlled independently via a control unit. Additionally, the present phases to transfer the design of a robot developed in a CAD tool to a virtual simulation environment without being lost the physical characteristics of the original design are showed, for which it is considered the various types of motions that the robot can perform depending on the design. Finally, shows the results obtained from the simulation of motion of a robot hexapod 18DOF and Theo Jansen mechanism.

On a method for the uniformization of solutions in central motion problems

The purpose of this study was to validate the accuracy of our novel 3D speckle tracking method by using numerical data, and to demonstrate the rapid processing of this method by using data obtained from human subjects. In order to create a method that can rapidly assess cardiac function in regional heart wall segments, a 3D speckle tracking algorithm was created that focuses on data quality and performance. Prototype application software based on this algorithm was written to evaluate cardiac wall motion and to calculate indices such as strain. Time series 3D image data were generated for artificial numerical models that simulate the shape of the left ventricle and exhibit various types of motion. We compared observed values returned by the algorithm with expected values yielded by the models. This software was also applied to volume data of the human heart acquired by a 3D ultrasound system. Measurement of the indices was evaluated by using an error ratio that was a residual between an expected value and estimated one divided by the expected value. The average error ratio of the time series volumes was less than 5% for all indices, and no individual error ratio was more than 10% for numerical models. The process was complete within 0.5s per frame for human heart volumes. This application software can provide good estimates of various wall motion indices. The results are similar to data from numerical models, and are provided quickly enough for routine clinical usage.

Availability analysis is an important issue in robotic swarm systems. It can help the designer to construct a cost-effective system with high availability and fewer resources. For the model and analysis to be fully specified and practical, this paper systematically investigates the major issues that need to be addressed in the analysis of various robotic swarm applications. Four models are established to consider systems with dependent and independent robots, homogenous and non-homogenous systems, and the effect of various types of motions on the overall system failure pattern. Detailed analysis of each of these models is performed based on renewal theory and continuous Markov Chain techniques. Numerical availability evaluations for two applications are also presented.

To describe the brake creep-groan phenomenon and several types of stick-slip motions, we propose both analytical and experimental investigations for an automatic transmission equipped vehicle. A lumped torsional model is employed to approximate the dynamics of the real mechanical system. This model assigns inertia to the drive, brake rotor, brake caliper and tire/vehicle, and by appropriate algorithms the simulation time histories include the effects of the friction non-linearity coupling brake and rotor. We consider how to force this particular system, from what physical state, and finding appropriate parameters and solutions. Important computational issues, as related to the stick-slip or slip-stick transitions, are addressed with the algorithms. Driving forces are assigned with two functions, one appropriate for comparison with the test and the other to study stick-slip orbits, which have been found to have various types of motion depending on the controlling brake actuation parameters. Since the groan features a discontinuous friction force and finite-repeating motions some comparisons are made in the frequency and time-frequency domains. These demonstrate the expected stick-slip frequency and multiple orders.

How Much the Eye Tells the Brain

We propose a motion generation model for simultaneous control of motion and force using deep learning. Conventional force control methods require expensive torque sensors and complex control theory, and implementing force control for each task requires huge development costs. In this paper, we realize rubbing motions against an uneven object at low cost by using a motion generation method that takes as input the joint angles and current values of an inexpensive servo motor. We evaluated the generalization ability of the model by confirming that the robot can perform rubbing motions against unlearned uneven or tilted objects. In addition, by comparing several motion generation models, we clarified that the following two components are important for simultaneous control of motion and force. (1) The joint angles and the current values are input to different neuron layers to extract the features. A time constant, which is the speed of information transfer, is set for each layer in order to integrate and learn input information with different time characteristics. (2) In the output part of the model, a single neuron layer is used to predict the joint angle and current value simultaneously. This makes it easy to extract features and integrate learning from two inputs with different time characteristics, and the robot can generate appropriate motions based on the contact situation in real time.

Multimodal feature extraction and fusion for audio-visual speech recognition