Silhouette-based gait recognition via deterministic learning

In this paper, we present a new silhouette-based gait recognition method via deterministic learning theory, which combines spatio-temporal motion characteristics and physical parameters of a human subject by analyzing shape parameters of the subject?s silhouette contour. It has been validated only in sequences with lateral view, recorded in laboratory conditions. The ratio of the silhouette?s height and width (H–W ratio), the width of the outer contour of the binarized silhouette, the silhouette area and the vertical coordinate of centroid of the outer contour are combined as gait features for recognition. They represent the dynamics of gait motion and can more effectively reflect the tiny variance between different gait patterns. The gait recognition approach consists of two phases: a training phase and a test phase. In the training phase, the gait dynamics underlying different individuals? gaits are locally accurately approximated by radial basis function (RBF) networks via deterministic learning theory. The obtained knowledge of approximated gait dynamics is stored in constant RBF networks. In the test phase, a bank of dynamical estimators is constructed for all the training gait patterns. The constant RBF networks obtained from the training phase are embedded in the estimators. By comparing the set of estimators with a test gait pattern, a set of recognition errors are generated, and the average L₁ norms of the errors are taken as the similarity measure between the dynamics of the training gait patterns and the dynamics of the test gait pattern. The test gait pattern similar to one of the training gait patterns can be recognized according to the smallest error principle. Finally, the recognition performance of the proposed algorithm is comparatively illustrated to take into consideration the published gait recognition approaches on the most well-known public gait databases: CASIA, CMU MoBo and TUM GAID.     

Efficient region-based motion segmentation for a video monitoring system

This paper presents an efficient region-based motion segmentation method for segmentation of moving objects in a traffic scene with a focus on a video monitoring system (VMS). The presented method consists of two phases: first, in the motion detection phase, the positions of moving objects in a scene are determined using an adaptive thresholding method. To detect varying regions by moving objects, instead of determining the threshold value manually, we use an adaptive thresholding method to automatically choose the threshold value. Second, in the motion segmentation phase, pixels that have similar intensity and motion information are segmented using a weighted k-means clustering algorithm to the binary region of the motion mask obtained in the motion detection. In this way, we need not process a whole image so computation time is reduced. Experimental results demonstrate robustness not only in the variation of luminance conditions and changes in environmental conditions, but also for occlusions among multiple moving objects.     

Effect of safety boots with toe spring on foot clearance features during walking

Tripping is a primary cause of occupational injury falls, especially among aging workers. This study investigated changes in foot clearance features during the normal walking swing phase affected by adding a toe spring height to safety boots. Gait data were obtained from nine male participants wearing experimental and control shoes via a motion capture system. A principal component analysis was conducted on three-dimensional foot segment trajectories and angles, and plantar surface motions were compared between conditions. Statistical analysis revealed significant principal component score differences between conditions in principal component vectors 1, 3, 5, 6, 8, and 9. The related swing phase three-dimensional plantar surface motion patterns were reconstructed. The results revealed two characteristics of the shoe-plantar surface three-dimensional motion of experimental shoes: higher foot clearance and a relatively straighter forward leg path. It was thus concluded that utilizing safety boots with toe springs may reduce falls in older workers due to occupational trips in industries.     

Synthesis of a stabilizing feedback for a wheeled robot with constrained control resource

Stabilization of motion of a wheeled robot with constrained control resource by means of a continuous feedback linearizing the closed-loop system in a neighborhood of the target path is considered. We pose the problem of finding the feedback coefficients such that the phase portrait of the nonlinear closed-loop system is topologically equivalent to that of a linear system with a stable node, with the asymptotic rate of decrease of the deviation from the target path being as high as possible. On this family, we pose the problem of minimization of “overshooting” for arbitrary initial conditions. The solution of this optimization problem is proved to be a limit discontinuous control law. A hybrid control law is proposed that, on the one hand, ensures the desired properties of the phase portrait and minimal overshooting and, on the other hand, does not result in a chattering inherent in systems with discontinuous feedbacks.     

Individual muscle contributions to the swing phase of gait: An EMG-based forward dynamics modelling approach

Muscle activation patterns and kinematic conditions at the beginning of the swing phase of gait were used as input to a forward dynamics simulation of the swing leg. A neuromusculoskeletal model was used to account for the non-linearity between muscle excitation and muscle force outputs. Following model tuning a close agreement between simulated and measured swing phase kinematics was obtained. Simulation results suggest that swing leg muscles play an important role in controlling the motion of the swing leg during walking, and that the effect of individual muscles is not necessarily restricted to the joints they span or their basic anatomical classifications.     

Representing short-term observations of moving objects by a simple visual language

In a variety of dynamical systems, formations of motion patterns occur. Observing colonies of animals, for instance, for the scientist it is not only of interest which kinds of formations these animals show, but also how they altogether move around. In order to analyse motion patterns for the purpose of making predictions, to describe the behaviour of systems, or to index databases of moving objects, methods are required for dealing with them. This becomes increasingly important since a number of technologies have been devised which allow objects precisely to get traced. However, the indeterminacy of spatial information in real world environments also requires techniques to approximate reasoning, for example, in order to compensate for small and unimportant distinctions which are due to noisy measurements. As a consequence, precise as well as coarse motion patterns have to be dealt with.A set of 16 atomic motion patterns is proposed. On the one hand, a relation algebra is defined on them. On the other hand, these 16 relations form the basis of a visual language using which motion patterns can easily be dealt with in a diagrammatic way. The relations are coarse but crisp and they allow imprecise knowledge about motion patterns to be dealt with, while their diagrammatic realisation also allow precise patterns to get handled. While almost all approaches consider motion patterns along arbitrary time intervals, this paper in particular focuses on short-term motion patterns as we permanently observe them in our everyday life.The bottom line of the current work, however, is yet more general. While it has been widely argued that it makes sense to use both sentential and diagrammatic representations in order to represent different things in the same system adequately (and hence differently), we argue that it makes even sense to represent the same things differently in order to grasp different aspects of one and the same object of interest from different viewpoints. We demonstrate this by providing both a sentential and a diagrammatic representation for the purpose of grasping different aspects of motion patterns. It shows that both representations complement each other.     

Nonsmooth dynamics of a 3D rigid body on a vibrating plate

In this work, we study a 3D rigid body, which is in a shape similar to a dumbbell, bouncing on a harmonically vibrating plate. The system involves two contact sets whose states vary with the relative motion between the plate and the dumbbell. These states are closely related to the physical properties of contact surfaces, and can be identified using the relationships of the relative kinematics. Under certain states, system-singular modes will likely occur due to the absence of the tangential compliance in Coulomb’s friction. Resolutions for these system singularities are given, and an integrated model, taking a hierarchical structure adaptable to the state variations, has been developed. Within an impact-free state, the contact forces to drive the motion of the dumbbell are obtained using an LCP (linear complementary problem) formulation. For the system in an impact state, the post-impact outputs to serve as the initial conditions for the subsequent motion can be calculated based on the new theory developed in impact dynamics. Specifying the dumbbell with initial states to bounce in a vertical plane, this model was justified by the comparison of our results and the experimental findings in other work. As certain periodic behaviors appear in the planar dynamics, the system in a 3D scenario also reveals intriguing patterns in the trajectories of the dumbbell’s mass center as projected onto the horizontal plane. They include a closed circular orbit, a dog-legged path, and a straight line.     

Development of a physiological knee motion simulator

Several kinds of knee motion simulator systems have been developed for the accurate analysis of knee biomechanics. Knee motion simulators, however, are not recognized for their practical use because of difficulties in design and control. In this study, a wire-driven knee simulator which generates physiological knee motion has been developed. Physiological three-dimensional tibia motion against the femur can be generated by the simulator. Many clinical studies have been performed to analyze the length displacement pattern of the anterior cruciate ligament (ACL) and the posterior cruciate ligament (PCL). We assume that the physiological knee motion can be realized if the length displacement patterns of the ACL and PCL against the knee flexion angle are the same as the experimental data obtained from the literature. A fuzzy neural control policy, one of the most effective intelligent control policies, has been applied for control of the simulator. Applying the fuzzy neural control policy, human knowledge and experience can be reflected and adaptive/learning ability can be incorporated in the controller. On-line learning of the fuzzy neural controller is carried out to minimize a fuzzy controlled evaluation function using the back-propagation learning algorithm. The effectiveness of the proposed simulator has been evaluated by experiments using a model bone.     

Development of a Complete Dynamic Model of a Planar Five-Link Biped and Sliding Mode Control of Its Locomotion During the Double Support Phase

This paper presents a complete dynamic model of a planar five-link biped walking on level ground. The single support phase (SSP), double support phase (DSP) and double impact occurring at the heel strike are included in the model. By modifying the conventional definition of certain physical parameters of the biped system, it is shown that the procedure of the derivation of the dynamic equations and their final forms are significantly simplified. For motion regulation during the DSP, our dynamic model is formulated as the motion of biped system under holonomic constraints, and the hip position and the trunk orientation are selected as the independent generalized coordinates to describe the constraint system and to eliminate the constraint forces from the equations of motion. Based on the presented dynamic formulation, we develop a sliding mode controller for motion regulation during the DSP where the biped is treated as a redundant manipulator. The stability and the robustness of the controller are investigated, and its effectiveness is demonstrated by computer simulations. To the best of our knowledge, it is the first time that a sliding mode controller is developed for biped walking during the DSP. This work makes it possible to provide robust sliding mode control to a full range of biped walking and to yield dexterity and versatility for performing specific gait patterns.     

Feasibility Study on Stability of Gait Patterns with Changeable Body Stiffness Using Pneumatic Actuators in a Quadruped Robot

An oscillator-type gait controller for a quadruped robot with antagonistic pairs of pneumatic actuators is proposed. By using the controller, a feasibility study on the stability of gait patterns with changeable body stiffness is reported. The periodic motions of the legs are generated and controlled by an oscillator network with state resetting. This type of controller has robustness in its gaits against variation in walking conditions or changes of environment. However, it sometimes loses robustness under conditions of actuation delay, decrease of actuator accuracy, etc. We investigated whether an oscillator-type controller with phase resetting is also effective under such conditions. The stability of locomotion also strongly depends on the mechanical properties of the body mechanism, especially the joint stiffness. In this report, the muscle tone of the robot on the pitching motion at the trunk is changeable by using the changeable elasticity of the pneumatic actuators. The stability of quadruped locomotion in walk and trot patterns with changeable body stiffness was evaluated with numerical simulations and hardware experiments.     

Reactive navigation of underwater mobile robot using ANFIS approach in a manifold manner

Learning and self-adaptation ability is highly required to be integrated in path planning algorithm for underwater robot during navigation through an unspecified underwater environment. High frequency oscillations during underwater motion are responsible for nonlinearities in dynamic behavior of underwater robot as well as uncertainties in hydrodynamic coefficients. Reactive behaviors of underwater robot are designed considering the position and orientation of both target and nearest obstacle from robot’s current position. Human like reasoning power and approximation based learning skill of neural based adaptive fuzzy inference system (ANFIS) has been found to be effective for underwater multivariable motion control. More than one ANFIS models are used here for achieving goal and obstacle avoidance while avoiding local minima situation in both horizontal and vertical plane of three dimensional workspace. An error gradient approach based on input-output training patterns for learning purpose has been promoted to spawn trajectory of underwater robot optimizing path length as well as time taken. The simulation and experimental results endorse sturdiness and viability of the proposed method in comparison with other navigational methodologies to negotiate with hectic conditions during motion of underwater mobile robot.     

Intentional motion on-line learning and prediction

Predicting motion of humans, animals and other objects which move according to internal plans is a challenging problem. Most existing approaches operate in two stages: (a) learning typical motion patterns by observing an environment and (b) predicting future motion on the basis of the learned patterns. In existing techniques, learning is performed off-line, hence, it is impossible to refine the existing knowledge on the basis of the new observations obtained during the prediction phase. We propose an approach which uses hidden Markov models (HMMs) to represent motion patterns. It is different from similar approaches because it is able to learn and predict in a concurrent fashion thanks to a novel approximate learning approach, based on the growing neural gas algorithm, which estimates both HMM parameters and structure. The found structure has the property of being a planar graph, thus enabling exact inference in linear time with respect to the number of states in the model. Our experiments demonstrate that the technique works in real-time, and is able to produce sound long-term predictions of people motion.     

Contrast thresholds and fixation disparity during 5-Hz sinusoidal single- and dual-axis (vertical and lateral) whole-body vibration

The present study assumed that whole-body vibration, transmitted through the seat, impairs spatial retinal resolution and oculomotor alignment parallel to the vibration axis. More specifically, that the decrement increases gradually from single-axis lateral via single-axis vertical and dual-axis linear to dual-axis circular motion. Twenty participants (19-26 years of age) with good vision volunteered for the experiment where in three sessions one of the following three conditions, contrast threshold, nonius bias or fixation disparity, for vertically and horizontally oriented test patterns was determined during five experimental conditions. The latter comprised a control (a(z) = a(y) = 0) and four conditions where 5-Hz sinusoidal motion of 1.2 ms(-2) rms were applied separately, either in the vertical or in the lateral direction, or simultaneously in both directions, once without and once with a phase shift of 90 degrees, thus causing dual-axis linear or circular motion. Contrast thresholds for horizontal gratings and the variability of vertical fixation disparity increased significantly whenever the participants were exposed to vertical motion (alone or combined with lateral motion). These effects may result in an increased difficulty in properly recognizing characters and graphic patterns containing horizontal lines and in the development of asthenopic complaints.     

Absolute stability of parametrically perturbed third-order systems

Consideration was given to the behavior of the third-order systems in phase space. Regularities of motion of the phase trajectories were established, and a criterion for absolute nonoscillation was obtained. For the absolutely nonoscillatory systems, the Hurwitz conditions serve as the absolute stability criterion. For the oscillatory systems, an additional Bulgakov condition was introduced to eliminate the possibility of parametric resonance. This condition which is verified on the invariant set defined using the Poincaré transform was shown to be a criterion for absolute stability of the oscillatory systems. The results obtained were used to solve the problem of absolute stability of a third-order control system with nonstationary sectorial nonlinearity.     

Robustness of time‐scale learning of robot motions to uncertainty in acquired knowledge

A disadvantage of present iterative learning control algorithms is that they are generally applicable only in cases where a certain task is performed over and over again. Consequently, if knowledge or control inputs acquired from learning a task can be used on similar tasks, learning will be more efficient. Recently, several methods for constructing the control input of a new motion based on the control inputs acquired from previous learning of similar tasks have been proposed. However, these methods assumed that the perfect control inputs could be obtained from the previous learning. In practice, the control inputs could never be obtained exactly from learning in the presence of certain uncertainties such as disturbance and measurement noises. In addition, it is also not known for sure how the basic motion patterns should be chosen for learning. In this article, the robustness problem of the time‐scale learning control to uncertainty in the acquired learning control inputs is formulated and solved. From the analysis, certain new insights such as its implication to choices of basic motion patterns for time‐scale learning will be discussed. Simulation results of a 3‐link robot are presented to illustrate the analysis. © 2001 John Wiley & Sons, Inc.     

Gender differences in lifting technique

The importance of different motion patterns in the assessment of work technique is rarely addressed in the literature and even less information can be found regarding gender differences. In this study the possible gender differences in lifting technique from lifting experiments on 12 female and 10 male participants were examined. The participants performed squat and stoop lifts of a box. Movements were measured by means of opto-electronic measurement systems. Kinematic data derived from the measurements revealed some differences between the men and the women, e.g. in trunk motion and knee angle ranges. The hip-knee interjoint coordination was more synchronized for women than for men in terms of the relative phase angle. It is concluded that so far gender differences in motion patterns have not been sufficiently explored and that men and women need to be considered separately in the evaluation of work technique in manual handling tasks. Advantages and disadvantages of different coordination patterns need to be further investigated.     

Gender differences in lifting technique

The importance of different motion patterns in the assessment of work technique is rarely addressed in the literature and even less information can be found regarding gender differences. In this study the possible gender differences in lifting technique from lifting experiments on 12 female and 10 male participants were examined. The participants performed squat and stoop lifts of a box. Movements were measured by means of opto-electronic measurement systems. Kinematic data derived from the measurements revealed some differences between the men and the women, e.g. in trunk motion and knee angle ranges. The hip-knee interjoint coordination was more synchronized for women than for men in terms of the relative phase angle. It is concluded that so far gender differences in motion patterns have not been sufficiently explored and that men and women need to be considered separately in the evaluation of work technique in manual handling tasks. Advantages and disadvantages of different coordination patterns need to be further investigated.     

A MATHEMATICAL MODEL THAT LEARNS AN ADAPTIVELY GENERATED NOVEL PATTERN IN QUADRUPED LOCOMOTION

Quadrupeds show several locomotion patterns when adapting to environmental conditions. An immediate transition among walk, trot, and gallop implies the existence of a memory for locomotion patterns. In this article, we postulate that motion pattern learning necessitates the repetitive presentation of the same environmental conditions and aim at constructing a mathematical model for new pattern learning. The model construction considers a decerebrate cat experiment in which only the left forelimb is driven at higher speed by a belt on a treadmill. A central pattern generator (CPG) model that qualitatively describes this decerebrate cat's behavior has already been proposed. In developing this model, we introduce a memory mechanism to store the locomotion pattern, where the memory is represented as the minimal point of the potential function. The recollection process is described as a gradient system of this potential function, while in the memorization process a new pattern learning is regarded as a new minimal point generation by the bifurcation from an already existing minimal point. Finally, we discuss the generalization of this model to motion adaptation and learning.     

Contrast thresholds and fixation disparity during 5-Hz sinusoidal single- and dual-axis (vertical and lateral) whole-body vibration

The present study assumed that whole-body vibration, transmitted through the seat, impairs spatial retinal resolution and oculomotor alignment parallel to the vibration axis. More specifically, that the decrement increases gradually from single-axis lateral via single-axis vertical and dual-axis linear to dual-axis circular motion. Twenty participants (19–26 years of age) with good vision volunteered for the experiment where in three sessions one of the following three conditions, contrast threshold, nonius bias or fixation disparity, for vertically and horizontally oriented test patterns was determined during five experimental conditions. The latter comprised a control (a _z= a _y= 0) and four conditions where 5-Hz sinusoidal motion of 1.2 ms^?2 rms were applied separately, either in the vertical or in the lateral direction, or simultaneously in both directions, once without and once with a phase shift of 90°, thus causing dual-axis linear or circular motion. Contrast thresholds for horizontal gratings and the variability of vertical fixation disparity increased significantly whenever the participants were exposed to vertical motion (alone or combined with lateral motion). These effects may result in an increased difficulty in properly recognizing characters and graphic patterns containing horizontal lines and in the development of asthenopic complaints.