Optimization-based posture reconstruction for digital human models

Digital human modeling provides a valuable tool for designers when implemented early in the design process. Motion capture experiments offer a means of validation of the digital human simulation models. However, there is a gap between the motion capture experiments and the simulation models, as the motion capture results are marker positions in Cartesian space and the simulation model is based on joint space. Therefore, it is necessary to map the motion capture data to simulation models by employing a posture reconstruction algorithm. Posture reconstruction is an inherently redundant problem where the collective distance error between experimental joint centers and simulation joint centers is minimized. This paper presents an optimization-based method for determining an accurate and efficient solution to the posture reconstruction problem. The procedure is used to recreate 120 experimental postures. For each posture, the algorithm minimizes the distance between the simulation model joint centers and the corresponding experimental subject joint centers which is called the mean measurement error.     

基于两段式IK与关键帧的虚拟人运动控制

为提高网络虚拟环境中虚拟人运动控制的效率与显示效果,提出逆运动学和关键帧法相结合的虚拟人运动控制方法.在逆运动学中设计改进的CCD算法实现分阶段IK计算,因回溯路径的缩短,减少了计算量;在关键帧中插入由四元数插值法生成的中间帧,提高虚拟人运动的真实感;通过IK求解的关节旋转角不仅实现对姿态的逼近,还使关节更具有弹性;引入运动特征参量,可增强虚拟人动作模式的多样性.     

Motion Compression using Principal Geodesics Analysis

Due to the growing need for large quantities of human animation data in the entertainment industry, it has become a necessity to compress motion capture sequences in order to ease their storage and transmission. We present a novel, lossy compression method for human motion data that exploits both temporal and spatial coherence. Given one motion, we first approximate the poses manifold using Principal Geodesics Analysis (PGA) in the configuration space of the skeleton. We then search this approximate manifold for poses matching end-effectors constraints using an iterative minimization algorithm that allows for real-time, data-driven inverse kinematics. The compression is achieved by only storing the approximate manifold parametrization along with the end-effectors and root joint trajectories, also compressed, in the output data. We recover poses using the IK algorithm given the end-effectors trajectories. Our experimental results show that considerable compression rates can be obtained using our method, with few reconstruction and perceptual errors.     

Human-like redundancy resolution: An integrated inverse kinematics scheme for anthropomorphic manipulators with radial elbow offset

As a mimic of the human arm structure, anthropomorphic manipulators with radial elbow offset (AMREO) are often deployed on humanoid service robots. However, the unique offset leads to difficulties in solving the analytical inverse kinematics (IK), which poses a challenge for further anthropomorphic control. This paper presents an integrated scheme for solving the path-wise IK problem of a 7-DoF AMREO in the position domain. Unlike other approaches, special attention is paid to the naturalness of the arm configuration, with the aim of making AMREO exhibit human-like behavior in human-centered environments. First, an analytical IK solution of AMREO for a single end-effector pose is derived based on the arm angle parameterization. Then, inspired by the habitual arm configurations in human reaching movements, the natural arm configuration mapped to wrist position is proposed for AMREO. To learn the patterns implied therein, a LSTM-based natural arm angle prediction network (NAPN) is designed and trained based on a human demonstration dataset. Finally, a redundancy resolution framework embedded with NAPN is built to generate smooth and natural joint configurations in the path-wise IK tasks. Comparative experiments show that the proposed analytical IK algorithm has better computational efficiency and precision than conventional methods, and can give complete results for one IK call within 4 μs. In addition, continuous path tracking experiments on a real robot validate the effectiveness and anthropomorphism of the redundancy resolution scheme based on NAPN.     

Optimization-based motion prediction of mechanical systems: sensitivity analysis

In this study, we derive sensitivity equations for the problem of optimization-based motion prediction of a mechanical system using the inverse recursive Lagrangian formulation. The simulation and sensitivity formulations are based on Denavit–Hartenberg transformation matrices. External forces and moments are taken into account in the formulation. The sensitivity information is needed in the optimization-based simulation process. The proposed formulation is demonstrated by calculating sensitivities for the optimal time trajectory planning problem of a two-link manipulator. In addition, sensitivities obtained using the proposed algorithm are compared to those obtained using the closed-form equations of motion. The two sensitivities match quite closely. The lifting motion of the two-link manipulator with external loads is also optimized by using the algorithm developed in this paper. More complex applications of the proposed formulation to digital human motion prediction are presented elsewhere.     

基于C3D数据的人体模型逆向动力学仿真分析方法

基于AnyBody人体建模仿真分析软件,采用试验得到的动作捕捉格式数据C3D文件,研究人体逆向动力学仿真分析方法,给出C3D驱动AnyBody人体运动仿真分析时的操作步骤、报错分析及其解决方案。仿真结果表明:人体模型基本参数设置、C3D参数设置、关键点的拟合和足底压力板参数调节是实现人体模型动态分析的关键,其能有效提高人体模型逆向动力学仿真精度。     

Backward walking simulation of humans using optimization

The objective of this study is to formulate, simulate and study the backward walking motion of a full-body skeletal digital human model using an optimization approach. Predictive dynamics is used to simulate the task in which joint angle profiles are treated as primary unknowns in the formulation. The joint torques are treated as dependent variables that are evaluated directly from the equations of motion. For the performance measure, the normalized dynamic effort represented by the integral of the squares of all the normalized joint torques is minimized subject to the associated physical constraints. Backward walking at different speeds is simulated and analyzed. The backward walking is validated with motion capture data and the available data in the literature. The results of the backward walking motion are compared to those of the forward walking motion in order to study the differences between the two walking patterns. It is seen that the joint torque profiles for hip and knee of backward walk are quite similar to those of forward walk with reverse sequence, but with different time duration of flexion and extension activations. These findings can impact many fields, such as improvement of human performance, rehabilitation from injuries, and others.     

仿人机器人复杂动作设计中人体运动数据提取及分析方法

提出了仿人机器人复杂动作设计中人体运动数据提取及分析方法. 首先, 通过运动捕捉系统获取人体运动数据, 并采用运动重定向技术, 输出人--机简化模型的数据; 然后, 对运动数据进行分析和运动学解算, 给出基于人体运动数据的仿人机器人逆运动学求解方法, 得到仿人机器人模型的关节角数据; 再经过运动学约束和稳定性调节后, 生成能够应用于仿人机器人的运动轨迹. 最终, 通过在仿人机器人BHR-2上进行刀术实验验证了该方法的有效性.     

Multi‐Variate Gaussian‐Based Inverse Kinematics

Inverse kinematics (IK) equations are usually solved through approximated linearizations or heuristics. These methods lead to character animations that are unnatural looking or unstable because they do not consider both the motion coherence and limits of human joints. In this paper, we present a method based on the formulation of multi‐variate Gaussian distribution models (MGDMs), which precisely specify the soft joint constraints of a kinematic skeleton. Each distribution model is described by a covariance matrix and a mean vector representing both the joint limits and the coherence of motion of different limbs. The MGDMs are automatically learned from the motion capture data in a fast and unsupervised process. When the character is animated or posed, a Gaussian process synthesizes a new MGDM for each different vector of target positions, and the corresponding objective function is solved with Jacobian‐based IK. This makes our method practical to use and easy to insert into pre‐existing animation pipelines. Compared with previous works, our method is more stable and more precise, while also satisfying the anatomical constraints of human limbs. Our method leads to natural and realistic results without sacrificing real‐time performance.     

Full-body performance animation with Sequential Inverse Kinematics

In this paper, we present an analytic-iterative Inverse Kinematics (IK) method, called Sequential IK (SIK), that reconstructs 3D human full-body movements in real time. The input data for the reconstruction is the least possible (i.e., the positions of wrists, ankles, head and pelvis) in order to be usable within a low-cost human motion capture system that would track only these six features. The performance of our approach is compared to other well-known IK methods in reconstruction quality and computation time obtaining satisfactory results for both. The paper first describes how we handle the spine and the clavicles before offering a simple joint limit model for ball-and-socket joints and a method to avoid self-collisions induced by the elbow. The second part focuses on the algorithms comparison study.     

Animating human locomotion with inverse dynamics

Because the major force components (the internal muscular forces and torques) are not known a priori over time, you cannot use forward dynamics to predict how the human body will walk. The alternative to the apparently intractable problem of specifying the joint torque patterns in advance is to use inverse dynamics to analyze the torques and forces required for the given motion. Such an analysis can show, for example, that the motion induces excessive torque, that the system is out of balance at a certain point, or that the step length is too great. We present a method of using an inverse dynamics computation to dynamically balance the resulting walking motion and to maintain the joint torques within a moderate range imposed by human strength limits. This method corrects or predicts a motion as indicated by the inverse dynamics analysis. Dynamic correctness is a sufficient condition for realistic motion of nonliving objects. In animating a self-actuated system, however, visual realism is another important, separate criterion for determining the success of a technique. Dynamic correctness is not a sufficient condition for this visual realism. An animation of dynamically balanced walking that is also comfortable in the sense of avoiding strength violations can still look quite different from normal human walking. A visually realistic and dynamically sound animation of human locomotion is obtained using an effective combination of kinematic and dynamic techniques     

A three-dimensional dynamic posture prediction model for in-vehicle seated reaching movements: development and validation

A three-dimensional dynamic posture prediction model for simulating in-vehicle reaching movements is presented. The model employs a four-segment 7-degrees-of-freedom linkage structure to represent the torso, clavicle and right extremity. It relies on an optimization-based differential inverse kinematics to estimate a set of four weighting parameters that quantify a timeconstant, inter-segment motion apportionment strategy. In the development 100 seated reaching movements performed by 10 subjects towards five in-vehicle targets were modelled, resulting in 100 sets of weighting Statistical analysis was then conducted to relate these parameters to and individual attributes. In the validation phase, the generalized model, parameter values statistically synthesized, was applied to novel data sets 700 different reaching movements (towards different targets and/or by subjects). The results demonstrated the model's ability to generate close in prediction: the overall mean time-averaged error in joint angle 5.2°, and the median was 4.7°, excluding reaches towards two extreme targets which modelling errors were excessive). The model's general success in and its unique characteristics led to implications with regard to the and underlying control strategies of human reaching movements.     

A three-dimensional dynamic posture prediction model for simulating in-vehicle seated reaching movements: development and validation

A three-dimensional dynamic posture prediction model for simulating in-vehicle seated reaching movements is presented. The model employs a four-segment 7-degrees-of-freedom linkage structure to represent the torso, clavicle and right upper extremity. It relies on an optimization-based differential inverse kinematics approach to estimate a set of four weighting parameters that quantify a time-constant, inter-segment motion apportionment strategy. In the development phase, 100 seated reaching movements performed by 10 subjects towards five typical in-vehicle targets were modelled, resulting in 100 sets of weighting parameters. Statistical analysis was then conducted to relate these parameters to target and individual attributes. In the validation phase, the generalized model, with parameter values statistically synthesized, was applied to novel data sets containing 700 different reaching movements (towards different targets and/or by different subjects). The results demonstrated the model's ability to generate close representations in prediction: the overall mean time-averaged error in joint angle was 5.2 degrees, and the median was 4.7 degrees, excluding reaches towards two extreme targets (for which modelling errors were excessive). The model's general success in prediction and its unique characteristics led to implications with regard to the performance and underlying control strategies of human reaching movements.     

基于足迹采样的运动编辑

针对人体两足动画提出一种基于足迹采样的运动编辑算法．足迹很好地描述了两足动画中必须满足的时空约束，通过调整足迹的位置与朝向来编辑两足动画是一种较为自然、直观的交互方式．为有效、快速地生成编辑后的动画，采用一种实时的逆向运动学算法求解两足动画中的支撑脚约束，然后使用层次B样条技术构造偏移映射完成编辑．为了便于逆向运动学算法的求解，提出基于采样的方法来计算质心轨迹．     

Learning to Walk by Imitation in Low-Dimensional Subspaces

In this paper, we provide the first demonstration that a humanoid robot can learn to walk directly by imitating a human gait obtained from motion capture (mocap) data without any prior information of its dynamics model. Programming a humanoid robot to perform an action (such as walking) that takes into account the robot's complex dynamics is a challenging problem. Traditional approaches typically require highly accurate prior knowledge of the robot's dynamics and environment in order to devise complex (and often brittle) control algorithms for generating a stable dynamic motion. Training using human mocap is an intuitive and flexible approach to programming a robot, but direct usage of mocap data usually results in dynamically unstable motion. Furthermore, optimization using high-dimensional mocap data in the humanoid full-body joint space is typically intractable. We propose a new approach to tractable imitation-based learning in humanoids without a robot's dynamic model. We represent kinematic information from human mocap in a low-dimensional subspace and map motor commands in this low-dimensional space to sensory feedback to learn a predictive dynamic model. This model is used within an optimization framework to estimate optimal motor commands that satisfy the initial kinematic constraints as best as possible while generating dynamically stable motion. We demonstrate the viability of our approach by providing examples of dynamically stable walking learned from mocap data using both a simulator and a real humanoid robot.     

Two‐handed human reach prediction models for ergonomic evaluation

As an essential function of computerized ergonomic evaluation models based on digital human models, realistic simulation or prediction of human reach profiles is of great importance. Although several human‐modeling efforts have been made to provide the capability of reach simulation, most studies have been limited to the reach of a single extremity. A variety of activities of human operators, however, frequently involve simultaneous positioning of two or more extremities to different target positions. Such a multiple reach problem cannot be satisfactorily resolved by means of conventional single‐extremity reach models because formulation of the problem as a series of single reaches rarely yields accurate trajectory of human‐reach profiles due to interactions of multiple extremities. In this research, a two‐handed reach prediction model was developed. The human upper body was modeled as a seven‐link system with 13 degrees of freedom, being regarded as a redundant open kinematic chain with two end‐effectors. As a way of solving the two‐handed reach problem, the resolved motion method was adopted among several inverse kinematics methods as the technique is fit for real‐time redundancy control. The method is also capable of incorporating the joint range availability criterion as a cost function to minimize excessive deviations of body joints from their neutral positions. Real human‐reach profiles were compared to those obtained from the prediction model and were found to be statistically similar. The methodology is expected to be applicable to the reach simulation of both upper and lower extremities without algorithmic difficulties. © 2010 Wiley Periodicals, Inc.     

虚拟人运动建模中的逆向运动学方法研究

针对当前虚拟人运动建模中缺少环境约束这一问题,文中提出了一种基于环境约束的逆向运动学求解方法。主要研究了在环境约束下,如何运用逆向运动学方法使人体的步行运动更加真实直观。首先提出了人体的层次结构模型和关节模型,其次对CCD算法的实现方法进行分析,最后结合CCD算法提出一种在环境约束下调整人体步行运动的方法。解决了运动捕捉方法中动作单一的问题,使虚拟人能够根据环境信息调整现有的动作,最终能够使虚拟人在平坦地面的运动和上下台阶运动都更加灵活真实。     

Kinematic analysis of 3SPS+1PS bionic parallel test platform for hip joint simulator based on unit quaternion

As a novel parallel hip joint simulator, the 3SPS+1PS bionic parallel test platform with 4 degrees of freedom including three rotations and one translation is proposed. SPS denotes the spherical-prismatic-spherical leg and PS denotes the prismatic-spherical leg where only the prismatic joint is actuated and hence underlined. By means of the unit quaternion method, the formulae for solving the inverse/forward displacement, the inverse/forward velocity and the inverse/forward acceleration kinematics are derived. Using the unit quaternion to represent the position and orientation of a moving platform, singularities caused by Euler angles can be avoided. Combining the topological structure characteristics of the 3SPS+1PS bionic parallel test platform and letting the three-dimensional (3-D) motion of a human hip joint as its output movement, the displacement trajectories of three active legs are constructed based on the inverse displacement kinematics. The forward kinematic tests whose data are recorded by a 3-D orientation capture system are carried out on the developed parallel hip joint simulator. Moreover, the results of the forward kinematic tests prove that the 3SPS+1PS bionic parallel test platform can approximately represent human hip joint motion and provide more reliable experimental data for hip joint prostheses in clinical application.     

Combining inverse blending and Jacobian‐based inverse kinematics to improve accuracy in human motion generation

We present in the paper a hybrid method for motion editing combining motion blending and Jacobian‐based inverse kinematics (IK). When the original constraints are changed, a blending‐based IK solver is first employed to find an adequate joint configuration coarsely. Using linear motion blending, this search corresponds to a gradient‐based minimization in the weight space. The found solution is then improved by a Jacobian‐based IK solver by further minimizing the distance between the end effectors and constraints. To accelerate the searching in the weight space, we introduce a weight map, which pre‐computes the good starting positions for the gradient‐based minimization. The advantages of our approach are threefold: first, more realistic motions can be generated by utilizing motion blending techniques, compared with pure Jacobian‐based IK. The blended results also increase the rate of convergence of the Jacobian‐based IK solver. Second, the Jacobian‐based IK solver modifies poses in the pose configuration space and the computational cost does not scale with the number of examples. Third, it is possible to extrapolate the given example motions with a Jacobian‐based IK solver, while it is generally difficult with pure blending‐based techniques. Copyright © 2014 John Wiley & Sons, Ltd.     

基于逆运动学和动力学的虚拟人行走仿真

在计算机上模拟真实人行走是计算机仿真的一个基本问题。人体行走是一种伴随着碰撞、摩擦和滑动的复杂的系统运动，为了实现模拟的逼真性，需要着重在运动控制上进行研究。首先对人体行走进行分析并建立了简单的人体模型，然后详细给出行走过程中关节点位置的数学描述，最后采用逆运动学求解雅可比矩阵的方法并结合动力学知识，运用VC＋＋．NET和OpenGL为编程工具以骨架模型实现了虚拟人行走。