Enriching a motion database by analogous combination of partial human motions

We have synthesized new human body motions from existing motion data, by dividing the body of an animated character into several parts, such as upper and lower body, and partitioning the motion of the character into corresponding partial motions. By combining different partial motions, we can generate new motion sequences. We select the most natural-looking combinations by analyzing the similarity of partial motions, using techniques such as motion segmentation, dimensionality reduction, and clustering. These new combinations can dramatically increase the size of a motion database, allowing more score in selecting motions to meet constraints, such as collision avoidance. We verify the naturalness and physical plausibility of the new motions using an SVM learning model and by analysis of static and dynamic balance.     

基于骨干长度比例之运动重定目标算法

为了减轻动画制作劳动强度，提高动画制作产能以及自动产生仿真动画，提出了一种结合运动捕获器（Motion Capture)技术与逆向运动学原理的用于完成关节动画中实时运动重定目标（Motion Retargeting)的新算法，该算法的主要概念是首先根据“原动者”与“标的者”之身材比例，推算出、标的者”末端效应器之定位，然后再利用逆向运动学之算法求得“标的者”各关节之旋转角度，因为该算法充分地利用了捕获器所纪录的“原动者”运动信息的密集重复性，而且所设计之定位法则能满足原运动对未端效应器之约束，所以该方法能展现出与原运动十分相似的动画，同时不违背原设定之约束，实验数据也展示，并说明了该算法之效能与优点。     

Style-Controllable Speech-Driven Gesture Synthesis Using Normalising Flows

Automatic synthesis of realistic gestures promises to transform the fields of animation, avatars and communicative agents. In off-line applications, novel tools can alter the role of an animator to that of a director, who provides only high-level input for the desired animation; a learned network then translates these instructions into an appropriate sequence of body poses. In interactive scenarios, systems for generating natural animations on the fly are key to achieving believable and relatable characters. In this paper we address some of the core issues towards these ends. By adapting a deep learning-based motion synthesis method called MoGlow, we propose a new generative model for generating state-of-the-art realistic speech-driven gesticulation. Owing to the probabilistic nature of the approach, our model can produce a battery of different, yet plausible, gestures given the same input speech signal. Just like humans, this gives a rich natural variation of motion. We additionally demonstrate the ability to exert directorial control over the output style, such as gesture level, speed, symmetry and spacial extent. Such control can be leveraged to convey a desired character personality or mood. We achieve all this without any manual annotation of the data. User studies evaluating upper-body gesticulation confirm that the generated motions are natural and well match the input speech. Our method scores above all prior systems and baselines on these measures, and comes close to the ratings of the original recorded motions. We furthermore find that we can accurately control gesticulation styles without unnecessarily compromising perceived naturalness. Finally, we also demonstrate an application of the same method to full-body gesticulation, including the synthesis of stepping motion and stance.     

基于多分辨率模型的快速运动过渡生成算法

运动捕获技术可以记录人体关节运动的细节,是当前最有前景的计算机动画技术.然而,运动数据的重用性一直是个难点,为此,多种运动编辑手段被提出.运动过渡是一种常见的编辑技术,它可以将输入的两端运动序列拼接,形成新的运动序列.其中,过渡点选择的合理与否直接影响着结果运动的质量.在两运动间选择过渡点,需要对输入运动的每一对帧之间分别计算帧间的距离,其计算复杂度是O(n2)的,通过引入多分辨率模型,文中将该复杂度降低到O(nlog2n),同时试验结果表明,此方法并未损害到结果运动的质量.     

基于视频的运动捕获

现有的运动捕获方法大都存在运动捕获设备昂贵、演员运动受限等缺点，为此，提出了一种利用视觉技术从视频中提取人体运动的方法，并对其中的特片跟踪和三维运动序列恢复等关键技术进行了深入研究。基于人体模型的特征跟踪算法利用卡尔曼滤波和极线方程，能精确地跟踪比较大的人体运动；采用不共面的非线性定标模型和考虑运动不确定性的三维重建方法，能恢复逼真的三维人体骨架模型，实验结果验证了基于视频的运动捕获方法的可行性和有效性。     

Detection and Synthesis of Full-Body Environment Interactions for Virtual Humans

We present a new methodology for enabling virtual humans to autonomously detect and perform complex full-body interactions with their environments. Given a parameterized walking controller and a set of motion-captured example interactions, our method is able to detect when interactions can occur and to coordinate the detected upper-body interaction with the walking controller in order to achieve full-body mobile interactions in similar situations. Our approach is based on learning spatial coordination features from the example motions and on associating body-environment proximity information to the body configurations of each performed action. Body configurations become the input to a regression system, which in turn is able to generate new interactions for different situations in similar environments. The regression model is capable of selecting, encoding and replicating key spatial strategies with respect to body coordination and management of environment constraints as well as determining the correct moment in time and space for starting an interaction. As a result, we obtain an interactive controller able to detect and synthesize coordinated full-body motions for a variety of complex interactions requiring body mobility. Our results achieve complex interactions, such as opening doors and drawing in a wide whiteboard. The presented approach introduces the concept of learning interaction coordination models that can be applied on top of any given walking controller. The obtained method is simple and flexible, it handles the detection of possible interactions and is suitable for real-time applications.     

Animated human agents with motion planning capability for 3D-space postural goals

In this paper, we present a rule-based heuristic method of motion planning for an animated human agent with massively redundant degrees of freedom. It constructs motion plans to achieve 3D-space goals of control points on the body, e.g. a hand, while avoiding collisions. Like the artificial potential field approach, the method performs motion decisions in 3D world space rather than in joint space. To handle the massively redundant degrees of freedom, we use a qualitative kinematic model, which specifies motions of body parts and dependencies among them, without specifying the exact distance parameters. This model helps the body select appropriate primitive motions for given goals of control points more globally than does the gradient vector of an artificial potential field of the body. The method simulates (in imagination) the suggested plan to find whether some body parts hit objects, and how much they penetrate the objects. Based on this simulated collision information, the method suggests intermediate goals of the collision body parts. A subplan to achieve these intermediate goals is again postulated by using the qualitative kinematic model. This explicit reasoning helps alleviate the basic cause of local minima in the potential field approach, namely, conflicts between attractive potential fields due to goal positions of control points and repulsive potential fields due to obstacles.     

Style Invariant Locomotion Classification for Character Control

We present a real‐time system for character control that relies on the classification of locomotive actions in skeletal motion capture data. Our method is both progress dependent and style invariant. Two deep neural networks are used to correlate body shape and implicit dynamics to locomotive types and their respective progress. In comparison to related work, our approach does not require a setup step and enables the user to act in a natural, unconstrained manner. Also, our method displays better performance than the related work in scenarios where the actor performs sharp changes in direction and highly stylized motions while maintaining at least as good performance in other scenarios. Our motivation is to enable character control of non‐bipedal characters in virtual production and live immersive experiences, where mannerisms in the actor's performance may be an issue for previous methods.     

The virtual stuntman: dynamic characters with a repertoire of autonomous motor skills

An ambitious goal in the area of physics-based computer animation is the creation of virtual actors that autonomously synthesize realistic human motions and possess a broad repertoire of lifelike motor skills. To this end, the control of dynamic, anthropomorphic figures subject to gravity and contact forces remains a difficult open problem. In this paper, we report on our ongoing development of a virtual stuntman, a dynamic graphical character that possesses a nontrivial repertoire of lifelike motor skills. The repertoire includes basic actions such as balance, protective stepping when balance is disturbed, protective arm reactions when falling, multiple ways of rising upright after a fall, and several more vigorously dynamic motor skills. Our virtual stuntman is the product of a recently proposed framework for integrating motor controllers, which includes among other ingredients an explicit model of pre-conditions; i.e., those regions of a dynamic figure's state space within which a given motor controller is applicable and expected to work properly.     

人体运动数据重构方法进展

运动捕捉系统产生的人体运动数据是标记点在运动序列中的位置数据,用于驱动人体模型产生真实感的动画。在对近几年有关人体运动数据重构的文献进行综合和分析的基础上,首先对人体运动数据重构进行了问题描述,并对人体运动数据在重构过程中难以避免的噪声问题和特征点(虚拟空间中的标记点)缺失问题的研究分别进行了总结和分析;然后对人体运动数据获取的光学式原型捕捉系统开发的研究进行了讨论,评述了人体运动数据驱动人体几何模型的相关研究;最后对未来研究提出了一些展望。     

Reconstructing 2D images with natural neighbour interpolation

In this paper, we explore image reconstruction by natural neighbour interpolation from irregularly spaced samples. We sample the image irregularly with techniques based on the Laplacian or the derivative in the direction of the gradient. Local coordinates based on the Voronoi diagram are used in natural neighbour interpolation to quantify the “neighbourliness” of data sites. Then we use natural neighbour interpolation in order to reconstruct the image. The main result is that the image quality is always very good in the case of the sampling techniques based on the Laplacian.     

Retrieval of logically relevant 3D human motions by Adaptive Feature Selection with Graded Relevance Feedback

A system that can retrieve logically relevant 3D captured motions is useful in game and animation production. We presented a robust logical relevance metric based on the relative distances among the joints. Existing methods select a universal subset of features for all kinds of queries which may not well characterize the variations in different queries. To break through this limitation we proposed an Adaptive Feature Selection (AFS) method that abstracts the characteristics of the query by a Linear Regression Model, and different feature subsets can be selected according to the properties of the specific query. With a Graded Relevance Feedback (GRF) algorithm, we refined the feature subset that enhances the retrieval performance according to the graded relevance of the feedback samples. With an ontology that predefines the logical relevance between motion classes in terms of graded relevance, the performance of the proposed AFS-GRF algorithm is evaluated and shown to outperform other class-specific feature selection and motion retrieval methods.     

Physical simulation of land vehicles with obstacle avoidance and various terrain interactions

Programming the motions of an autonomous planetary robot moving in an hostile and hazardous environment is a complex task which requires both the construction of nominal motion plans and the anticipation as far as possible of the effects of the interactions existing between the vehicle and the terrain. In this paper we show how physical models and dynamic simulation tools can be used for amending and completing a nominal motion plan provided by a classical geometrical path planner. The purpose of our physical modeller-simulator is to anticipate the dynamic behaviour of the vehicle while executing the nominal motion plan. Then the obtained simulation results can be used to assess and optimize the nominal motion plan. In the first part, we outline the physical models that have been used for modelling the different types of vehicle, of terrain and of vehicle-surface interactions. Then we formulate the motion planning problem through the definition of two basic abstract constructions derived from physical model: the concept of generalized obstacle and the concept of physical target. We show with various examples how it is possible, when using this method, to solve the locomotion problem and the obstacle avoidance problem simultaneously and, furthermore, to provide the human operator with a true force feedback gestural control over the simulated robot.     

Skill-level based examination of forearm muscle activation associated with efficient wrist and finger movements during typing

This study aims to elucidate the relationship between the wrist and finger movements and forearm muscle activation of twelve young people (age: 21.1 ± 0.76 years, nine men and three women) with different typing skills. We hypothesize that skilled typists (STs) could move their wrist and finger joints faster than unskilled typists (UTs) because they could efficiently use their muscles according to the activity characteristics of the flexors and extensors of the wrist joint. We measured wrist and finger movements using a 3D motion capture system and forearm muscle activation using surface electromyography during the typing task. We analyzed the entire task and the time when the U key was entered during the same task. The angular velocity of the wrist and finger flexion/extension and the muscle activation of the wrist flexors was higher in the STs than in the UTs, while the muscle activation of the wrist extensors was higher in the latter than that in the former. Our results showed that STs may have used their forearm muscles to take advantage of the physical characteristics of the keys and the spring characteristics of their muscles and tendons. It was suggested that they placed less mechanical stress on their finger muscles and tendons when pressing and releasing the keys.     

Markerless Multiview Motion Capture with 3D Shape Model Adaptation

In this paper, we address simultaneous markerless motion and shape capture from 3D input meshes of partial views onto a moving subject. We exploit a computer graphics model based on kinematic skinning as template tracking model. This template model consists of vertices, joints and skinning weights learned a priori from registered full‐body scans, representing true human shape and kinematics‐based shape deformations. Two data‐driven priors are used together with a set of constraints and cues for setting up sufficient correspondences. A Gaussian mixture model‐based pose prior of successive joint configurations is learned to soft‐constrain the attainable pose space to plausible human poses. To make the shape adaptation robust to outliers and non‐visible surface regions and to guide the shape adaptation towards realistically appearing human shapes, we use a mesh‐Laplacian‐based shape prior. Both priors are learned/extracted from the training set of the template model learning phase. The output is a model adapted to the captured subject with respect to shape and kinematic skeleton as well as the animation parameters to resemble the observed movements. With example applications, we demonstrate the benefit of such footage. Experimental evaluations on publicly available datasets show the achieved natural appearance and accuracy.     

Dynamic output feedback robust MPC with convex optimisation for system with polytopic uncertainty

This paper proposes a dynamic output feedback robust model predictive control for system with polytopic model uncertainty. Earlier researches on this topic utilise iterative methods to solve the non-convex optimisation problem which are computationally demanding. In order to reduce the computational burden, we explore a new approach in this paper, where, by utilising some proper matrix transformations, a computationally more efficient but conservative convex optimisation problem is formulated which can be solved in terms of linear matrix inequalities. Furthermore, we try to reduce the conservativeness by introducing a nonsingular matrix as a degree of freedom. The recursive feasibility and the convergence of the augmented state to the equilibrium point are guaranteed. The effectiveness of the proposed approach is illustrated by two numerical examples.     

Pitch Angle Control of Unmanned Air Vehicle with Uncertain System Parameters

In this paper a new algorithm of trajectory tracking based on closest radius solution of the interval equations system is proposed. The design procedure is given and applied to the pitch angle control of unmanned testing rocket with uncertain parameters. The proposed algorithm gives a framework to design a control for a wide range of different linear time-invariant processes with uncertain parameters and can be implemented also in the case of non-convex problems. The algorithm gives the analytical way of finding the nearly optimal solution of model reference trajectory tracking in the case of general time-invariant systems with uncertain parameters and can be used when optimization method fails due to the complexity of the problem.     

舰载机着舰侧回路时变风险权值矩阵线性变参数预测控制

针对舰载机钩索阻拦阶段,提出阻拦风险函数的概念和计算方法,通过离线计算获得时变状态权值矩阵和时变控制输入权值矩阵.采用时变状态权值矩阵可以实时调整各状态之间的变化关系,而时变的控制输入权值矩阵能够调整副翼和方向舵的输入峰值,避免出现输入饱和情况,提高横侧向自动着舰的效率和安全性.建立了时变的系统输出约束函数,增加了舰载机进舰阶段控制器的可解性.建立了基于状态偏差的舰载机横侧向着舰模型,针对舰载机上很多状态不能直接测量的特点,采用离线设计状态观测器的方法来估计舰载机的真实状态值,利用线性矩阵不等式求解系统最优解,通过舰载机自动着舰三维仿真模拟平台验证了算法的可行性.     

Tracking control of electro-hydraulic servo multi-closed-chain mechanisms with the use of an approximate nonlinear internal model

This paper studied the trajectory-tracking problem of a hydraulic servo multi-closed-chain mechanism. The nonaffine nonlinear characteristic of the electro-hydraulic actuator and its time-varying uncertainty load resulting from the multi-closed-chain mechanism was taken into consideration in the proposed novel nonlinear control algorithm, that is, the approximate internal model control (AIMC) integrated with a position feedback control in cascade control design. This algorithm improves the trajectory-tracking performance of the hydraulic servomechanism (HSM). To reduce the difficulty in directly utilizing the AIMC for the HSM position trajectory, the complex electro-hydraulic mechanical system was divided into two subsystems: nonaffine nonlinear, and linear. The AIMC controller was designed for the nonaffine nonlinear subsystem to realize velocity trajectory tracking control, whereas a position feedback control was derived for the linear subsystem. The position trajectory tracking control was achieved by congruently combining the AIMC, and the position feedback control based on a recursive design idea. In addition, a complete state-space mathematical model for the HSM was developed and illustrated through simulations and experiments. Based on the proposed approach and the AIMC, the desired position and velocity trajectory tracking was examined on a hydraulic forging manipulator. The stability of the proposed method was analytically derived. Results of the simulations and experiments performed with the hydraulic manipulator demonstrated the effectiveness of the proposed approach.     

Robust model predictive control with guaranteed setpoint tracking

In this paper a novel robust model predictive control (RMPC) algorithm is proposed, which is guaranteed to stabilize any linear time-varying system in a given convex uncertainty region while respecting state and input constraints. Moreover, unlike most existing RMPC algorithms, the proposed algorithm is guaranteed to remove steady-state offset in the controlled variables for setpoints (possibly) different from the origin when the system is unknown linear time-invariant. The controller uses a dual-mode paradigm (linear control law plus free control moves to reach an appropriate invariant region), and the key step is the design of a robust linear state feedback controller with integral action and the construction of an appropriate polyhedral invariant region in which this controller is guaranteed to satisfy the process constraints. The proposed algorithm is efficient since the on-line implementation only requires one to solve a convex quadratic program with a number of decision variables that scale linearly with the control horizon. The main features of the new control algorithm are illustrated through an example of the temperature control of an open-loop unstable continuous stirred tank reactor.