Existing Motion Data Research Articles

Scenario-Based Sensed Human Motion Editing and Validation Through the Motion-Sphere

Synthesizing realistic human motion data using a real-time motion capture system in a controlled environment is a critical challenge. In addition, effectively manipulating the existing motion data is another primary concern and using such modified data in human motion analysis and activity recognition systems are prone to errors. This paper presents a simplistic and comprehensive system to effortlessly author, edit, and validate human motion data. The system enables a naive user to edit the existing motion data interactively using a humanoid model in a three-dimensional space, based on user-defined scenarios and synthesize numerous variations of the motion sequences. A modular concept of scenario-based sensed unit motion editing has been adopted to demonstrate the proposed system. We employed an efficient analytical kinematic and constraint solver to enforce the inherent body joint limits and external constraints while editing to synthesize complete and meaningful motion sequences. Furthermore, we substantiated the proposed sensed unit motion editing framework through a visual validation study using an open-source intuitive visualization tool called the Motion-Sphere. Finally, we compared the resultant synthesized motion against the real-time motion capture system data to verify the body segments’ orientation and position accuracy deviations.

변형 자동 인코더를 활용한 모션 스타일 이전

In this paper, we propose a framework that transfers the information of style motions to content motions based on a variational autoencoder network combined with a style encoding in the latent space. Because we transfer a style to a content motion that is sampled from a variational autoencoder, we can increase the diversity of existing motion data. In addition, we can improve the unnatural motions caused by decoding a new latent variable from style transfer. That improvement was achieved by additionally using the velocity information of motions when generating next frames.

Precurved, Fiber-Reinforced Actuators Enable Pneumatically Efficient Replication of Complex Biological Motions.

Much of the research on bioinspired soft robotics has focused on capturing the interplay of biological form and function. However, existing soft robotic actuators are mostly made with linear or planar fabrication orientations that do not represent the resting geometry of complex biological systems, such as curved musculature. This work introduces the ability to create fiber-reinforced actuators with precurved configurations. By tuning variables such as dimensions and fiber angles, an optimization algorithm can prescribe the mechanical fabrication parameters to create a fiber-reinforced actuator that can generate controlled motion to follow a desired input trajectory. Precurved configurations introduce an additional optimization parameter, the initial bend angle, allowing for a more accurate and robust algorithm and generating a median percent error of <1%. With a customized software tool, we can take existing motion data from biological systems-such as medical imaging-and build soft robotic actuators optimized to replicate these trajectories. We can predict the motion of precurved actuators both analytically and numerically and replicate the motion experimentally, with excellent trajectory matching between the three. In constructing actuators that better match the native forms found within biological systems, we find that precurved actuators are more efficient than their initially straight counterparts. This pneumatic efficiency allows for the use of control systems with lower power and precision, lowering the economic cost of the associated control hardware, while more accurately replicating the biological motion. Taking two examples from biology, that of the human diaphragm during respiration and that of a jellyfish bell during locomotion, we design and generate fiber reinforced actuators to mimic these motions.

Optimizing Neural Network as Locomotion Controller With Motion Data

This paper proposes a novel method of optimizing locomotion controller for virtual characters, which is constructed with the deep neural network and learned with the assistance of existing motion data. The learning is accomplished with a progressive reward function by starting from a simple reward function and gradually imposing advanced requirements on the gait pattern. A multi-critic model is proposed to address the dynamically changing reward function by evaluating the individual goal in a separate fashion. This strategy proves effective in avoiding the local minima that are likely to occur when a set of weighted reward functions are introduced at the beginning of the optimization. The results show that the integration of motion data not only ensures the consistency between the synthetic and original motions but also accelerates the learning of network parameters. We demonstrate the application of our method to a variety of virtual characters (cheetah, hopper, 2-D walker, and 3-D humanoid) performing various tasks (walking, running, jumping, and traversing at different velocities and across uneven terrains).

Dimensionality Reduction and Generation of Human Motion

To reuse existing motion data and generate new motion, a method of human motion nonlinear dimensionality reduction and generation, based on fast adaptive scaled Gaussian process latent variable models, is proposed. Through statistical learning on motion data, the motion data are mapped from high-dimensional observation space to low-dimensional latent space to implement nonlinear dimensionality reduction, and probability distributing of posture space which measures the nature of posture is obtained. The posture which meets constraints and has maximal probability can be computed as the solution of inverse kinematics. This method can avoid cockamamie computation and posture distortion existing in traditional inverse kinematics.The experiments show that our method has higher convergence velocity and precision and extends editing range of motion by adapting motion editing direction.

3d human motion retrieval based on human hierarchical index structure.

With the development and wide application of motion capture technology, the captured motion data sets are becoming larger and larger. For this reason, an efficient retrieval method for the motion database is very important. The retrieval method needs an appropriate indexing scheme and an effective similarity measure that can organize the existing motion data well. In this paper, we represent a human motion hierarchical index structure and adopt a nonlinear method to segment motion sequences. Based on this, we extract motion patterns and then we employ a fast similarity measure algorithm for motion pattern similarity computation to efficiently retrieve motion sequences. The experiment results show that the approach proposed in our paper is effective and efficient.

Physically Based Motion Adaptation for Diverse Articulated Figures

In this paper, we propose a new physically based method to adjust the original motion to meet adaptation requirements when animating 3D diverse virtual characters. We refer to an input motion sequence and user specifications; subsequently, the proposed method calculates physically consistent adjustments to fill the adaptation requirements. Motion constraints lead movement of virtual characters to follow the laws of physics, and we refine motion data with a series of minimal adjustments in each step of iterative procedures. Since the input motion must be a physically consistent motion, we apply the numerical method to perform only a few slight modifications in order to ensure the new motion that will be close to the original motion with nature and smooth features. Moreover, our method emphasizes the reuse of the existing motion data. Users arbitrarily set up their virtual environments, make some changes on the character’s skeletal structure, and specify adaptation requirements through a GUI interface. The system will automatically modify the input motion and then generate the new desired motion. The experimental results show that the proposed method meets the kinematic, dynamic, and environment constraints to produce the adapted motion towards different characters.

Decomposing biological motion: A linear model for analysis and synthesis of human gait patterns

We present an algorithm that transforms visual motion data such that they can be successfully approached with linear methods from statistics and pattern recognition. The transformation is based on a linear decomposition of postural data into a few components that change with sinusoidal temporal patterns. The components repeat consistently across subjects such that linear combinations of existing motion data result in smooth, meaningful interpolations.

1P1-J18 大規模モーションデータベースの構築とアニメーション制作支援への応用(デジタルヒューマン)

We propose a large-scale motion database structure for searching and reusing existing motion data. Motion data from different sources, marker sets, and skeleton models can be included in the database. All samples from multiple motion clips are put into a single database to consider transition from one clip to another. The database has a binary-tree structure to enable fast search from key motions. We demonstrate the usability of the database through an application to assisting keyframing in character animation.

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.

Real‐time locomotion control by sensing gloves

AbstractSensing gloves are often used as an input device for virtual 3D games. We propose a new method to control characters such as humans or animals in real‐time by using sensing gloves. Based on existing motion data of the body, a new method to map the hand motion of the user to the locomotion of 3D characters in real‐time is proposed. The method was applied to control locomotion of characters such as humans or dogs. Various motions such as trotting, running, hopping, and turning could be produced. As the computational cost needed for our method is low, the response of the system is short enough to satisfy the real‐time requirements that are essential to be used for games. Using our method, users can directly control their characters intuitively and precisely than previous controlling devices such as mouse, keyboards or joysticks. Copyright © 2006 John Wiley &amp; Sons, Ltd.

다양한 지형에서의 걷기와 달리기 동작의 효율적 생성

In 3D animation most people adjust the moving motion of their characters on various terrains by using motion data acquired with the motion capture equipment. The motion data can be used to present real human motions naturally, but the data must be captured again to apply to the different terrains from those given at acquiring mode. In addition, there would be a difficulty when applying the data to other characters, in that case the motion data must be captured newly or the existing motion data must be heavily edited manually. In this paper we propose a unified method to generate human motions of walking and running for various terrains such as flat plane, inclined plane, stairway and irregular face. With these methods we are able to generate human motions controlled by the parameters : body height, moving speed, stride, etc. In the proposed methods, the positions and angles of joint can be calculated by using inverse kinematics, and we calculate the trajectory of the swing leg and pelvis according to the cubic spline. With these methods we were presented moving motions using a model of a human body.

Real-time interactive motion transitions by a uniform posture map

Motion transition is a useful method for reusing existing motion data. It generates a seamless intermediate motion with two short motion sequences. In this paper, the Uniform Posture Map (UPM) is proposed to perform motion transitions. The UPM is organized through the quantization of various postures with an unsupervised learning algorithm; it places the output neurons with similar postures in adjacent positions. Using this property, an intermediate posture of two applied postures is generated; the generating posture is used as a key-frame to make an interpolating motion. The UPM has lower computational costs in comparison with other motion transition algorithms. It provides a control parameter; an animator can not only control the motion simply by adjusting this parameter, but can also produce animation interactively. The UPM prevents the generating of invalid output neurons to present unrealistic postures in the learning phase; thus, it makes more realistic motion curves. Finally, it contributes to the making of more natural motions. The motion transition algorithm proposed in this paper can be applied to various fields such as real-time 3D games, virtual reality applications, and web 3D applications.

효율적인 애니메이션을 위한 모션 에디팅 방법의 계산량분석에 관한 연구

Regarding efficient development of computer graphic animations, lots of techniques for editing or transforming existing motion data have been developed. Basically, the motion transformation techniques follow optimization process. To make the animation be natural, almost all the techniques utilize kinematics and dynamics in constructing constraints for the optimization. Since the kinematic and dynamic structures of virtual characters to be animated are very complex, the most time-consuming part is known to the optimization process. In order to suggest some guide lines to engineers involved in the motion transformation, in this paper, we analyze the computational complexities for typical motion transformation in quantitative manner as well as the possibility for parallel computation.

The creative process of animating human movement

The computer animation of articulated models such as human figures can be regarded as a complex synthesis task much like composing a musical piece or designing a new product. Developing computer animation tools requires a sound understanding of both this creative process as well as the animation task itself. Most current commercial animation packages focus on providing useful but low-level functions to create, modify and compose human movement. This paper proposes a hierarchical approach to human figure animation which incorporates knowledge on several levels about movement and motion processing in order to support the creative process and facilitate the motion specification task without taking control away from the animator. Two examples are given to illustrate this principle: a tool for animation of human locomotion, and motion signal processing techniques to adapt existing motion data.