Analysis and Clustering of Upper Limb Motion during the Hand Dexterity Pegboard Test using Inertial Sensor Systems.

Maintaining hand and upper limb mobility is important from the viewpoint of freedom in daily life and high performance in work. Few studies on the mobility and dexterity of the upper limb have focused on detailed hand and finger movements. Therefore, we measured the motion of the upper limbs during a general hand dexterity pegboard test using inertial sensor systems and our previous measuring method. To clarify the characteristics of each purpose of motion, we divided the peg-in-hole motion in the pegboard test into its three sections, focusing on two sections: the pinch section, and the carry and insert section. In addition, the obtained joint angles were grouped into arm group and finger group, and singular value decomposition was performed for each joint group in each section. By clustering the decomposition results across five subjects' multiple right and left arm tests, and averaging the singular vectors in the same cluster, the joint distributions and combinations could be clarified. In addition, by recalculating joint angles from averaged SVD results and applying them to the rigid link model, we obtained motion animation with characteristics that made it possible to more clearly understand the requirements for greater dexterity. These results suggested high dexterity motion characteristics in the pinch section, and the carry and insert section of the pegboard test.

Analysis of Dexterity Motion by Singular Value Decomposition for Hand Movement Measured Using Inertial Sensors.

Finger movements play an important role in many daily human actions. Among the studies on the dexterity of fingers required for various tasks in neurology and simple evaluation tests, few have focused on detailed finger movements themselves. Therefore, in this study, we improved the hand motion measurement system using inertial sensors and the motion analysis method developed in our previous study and measured the motion of the upper limbs (including the fingers) during a general finger dexterity test. By applying singular value decomposition to the obtained joint angles and decomposing them into simpler movement units, we obtained the timing of each movement unit and the purpose of each movement as the coordination state of the joints. By applying hierarchical clustering to multiple trials in a finger dexterity test, we also determined the similarity between trials and investigated the characteristics of movements with higher dexterity. We investigated the motor characteristics in finger dexterity by analyzing our results.

Time-series changes in intramuscular coherence associated with split-belt treadmill adaptation in humans.

Humans can flexibly modify their walking patterns. A split-belt treadmill has been widely used to study locomotor adaptation. Although previous studies have examined in detail the time-series changes in the spatiotemporal characteristics of walking during and after split-belt walking, it is not clear how intramuscular coherence changes during and after split-belt walking. We thus investigated the time-series changes of intramuscular coherence in the ankle dorsiflexor muscle associated with split-belt locomotor adaptation by coherence analysis using paired electromyography (EMG) signals. Twelve healthy males walked on a split-belt treadmill. Surface EMG signals were recorded from two parts of the tibialis anterior (TA) muscle in both legs to calculate intramuscular coherence. Each area of intramuscular coherence in the beta and gamma bands in the slow leg gradually decreased during split-belt walking. Significant differences in the area were observed from 7 min compared to the first minute after the start of split-belt walking. Meanwhile, the area of coherence in both beta and gamma bands in the fast leg for the first minute of normal walking following split-belt walking was significantly increased compared with normal walking before split-belt walking, and then immediately returned to the normal walking level. These results suggest that cortical involvement in TA muscle activity gradually weakens when adapting from a normal walking pattern to a new walking pattern. On the other hand, when re-adapting from the newly adapted walking pattern to the normal walking pattern, cortical involvement might strengthen temporally and then weaken quickly.

Variant and Invariant Spatiotemporal Structures in Kinematic Coordination to Regulate Speed During Walking and Running.

Humans walk, run, and change their speed in accordance with circumstances. These gaits are rhythmic motions generated by multi-articulated movements, which have specific spatiotemporal patterns. The kinematic characteristics depend on the gait and speed. In this study, we focused on the kinematic coordination of locomotor behavior to clarify the underlying mechanism for the effect of speed on the spatiotemporal kinematic patterns for each gait. In particular, we used seven elevation angles for the whole-body motion and separated the measured data into different phases depending on the foot-contact condition, that is, single-stance phase, double-stance phase, and flight phase, which have different physical constraints during locomotion. We extracted the spatiotemporal kinematic coordination patterns with singular value decomposition and investigated the effect of speed on the coordination patterns. Our results showed that most of the whole-body motion could be explained by only two sets of temporal and spatial coordination patterns in each phase. Furthermore, the temporal coordination patterns were invariant for different speeds, while the spatial coordination patterns varied. These findings will improve our understanding of human adaptation mechanisms to tune locomotor behavior for changing speed.

Hand Motion Measurement using Inertial Sensor System and Accurate Improvement by Extended Kalman Filter.

Analysis of hand motions is crucial in such actual conditions as daily life and traditional work. We developed a measurement system using inertial sensors instead of an optical motion capture system that measures with spatial constraints. However, for these sensors, the posture error caused by the integration of the angular velocity is critical. A typical solution uses sensor fusion with simple observation equations to measure such lower limbs by walking analysis. For finger motions, a simple observation, calculated identically as the initial posture, is unsuitable because fingers may be moved intricately and quickly by multiple joints and parallel links. Therefore, we constructed an observation equation based on such dynamic acceleration as rotational acceleration and the correction of compass error. Using this suggested observation equation, since both the posture and position error were verified in the hand and forearm motions by a comparison with the optical motion capture, we could measure them with high accuracy. After measuring the movements of an actual hand, such as writing words and spinning a top, we analyzed the characteristics from a reproduced link model and joint angles.

Modeling of Hand and Forearm Link using Inertial Sensors.

In this paper, we describe our hand and forearm motion measuring system using inertial sensors. Although a sensor axis is often used to measure human body motion, it is prone to attachment and offset error, and, because a hand is made of five parallel links, it is not suitable for hand measurement. Therefore, we propose a method of modeling the hand and forearm link using only sensors. To model a finger link, a compass' azimuth deviation is important because it may affect the intersection of the fingers. Therefore, before modeling, we applied correction and alternative methods using the angular velocity direction. As a result, the deviation between the sensors decreased to 1/5. During the modeling, we estimated the sensor's position vectors on the distal and next distal segments from each joint, and calculated the relationship matrices between the sensors through initial posture during estimation. We created the hand and forearm link model by combining the position vectors and relationship matrices. Comparison with optical motion capture showed that the hand shape without intersection of the fingers matched well, but there was offset because the forearm sensors deviated from the estimated positions. Although we must improve the attachment to the forearm, these methods proved effective for hand motion measurement.

Estimation of joint center and measurement of finger motion by inertial sensors.

The authors developed a wearable finger motion measurement system using inertial and geomagnetic sensors. Using this system, motion and posture of the hands and fingers can be measured. However, the joint center and segment axis cannot be accurately measured in a previous study using the sensors. Therefore, the authors proposed a method of estimating the joint center and segment lengths. This method utilizes the fact that the calculation formula of the rotational acceleration in the sensor's coordinate system is the vector product. However, because the vector product is irreversible, the rotation center was calculated by using a position vector placed on an intersection line of two planes constituted by rotational acceleration. As a result of the verification, estimation error was small. In addition, finger motion was measured using posture measurement and rotation center estimation. This measured motion, joint angle and segment lengths estimated by the finger motion measurement system were compared with a motion capture system. As a result, the initial MP and PIP joint angles had some differences, but there was no difference for the MP and PIP joint angles during movement and the other joint angles. Therefore, we need to improve the estimation method of the PIP and DIP joint centers.

Real-time motion discrimination considering variation of EMG signals associated with lapse of time.

This study proposes a motion discrimination method that considers the variation of electromyogram (EMG) signals associated with a lapse of time. In a previous study, we proposed a real-time discrimination method based on EMG signals of the forearm. Our method uses a hypersphere model as a discriminator. In motion discrimination using EMG signals, one problem is to maintain high discrimination accuracy over time because EMG signals change with a lapse of time. This study analyzed the effect of changes in EMG signals on our method. Based on analysis results, adding a relearning system of the decision criteria to the discrimination system was expected to be effective. We created a new motion discrimination method that contains the relearning system and experimentally verified its effectiveness. The motion discrimination system discriminated three hand motions, open, grasp, and pinch with discrimination accuracy above 90% in real-time (processing time below 300 ms) even after time elapsed.

Motion discrimination technique by EMG signals using hyper-sphere model.

This study developed a method of discriminating real-time motion from electromyogram (EMG) signals. We previously proposed a real-time motion discrimination method using hyper-sphere models that discriminated five motions (open, grasp, pinching, wrist extension, and wrist flexion) above 90% and quickly learned EMG signals. Our method prevents elbow motions from interfering with hand motion discrimination. However, we presume in our method that feature quantities do not change with time. Discrimination accuracy might deteriorate over time. Additionally, our method only discriminated three motions (open, grasp, pinching) for finger motions. This paper proposes the effectiveness of our method for changing feature quantities caused by time variation and a real-time motion discrimination method using new hyper-sphere models for four finger motions (open, grasp, pinching, and 2-5th finger flexion). We carried out two experiments and verified the effectiveness of our method for changing feature quantities and four finger motions discrimination using the new hyper-sphere models.

Quantitative evaluation of unrestrained human gait on change in walking velocity.

In human gait motion analysis, which is one useful method for efficient physical rehabilitation to define various quantitative evaluation indices, ground reaction force, joint angle and joint loads are measured during gait. To obtain these data as unrestrained gait measurement, a novel gait motion analysis system using mobile force plates and attitude sensors has been developed. On the other hand, a human maintains a high correlation among the motion of all joints during gait. The analysis of the correlation in the recorded joint motion extracts a few simultaneously activating segmental coordination patterns, and the structure of the intersegmental coordination is attracting attention to an expected relationship with a control strategy. However, when the evaluation method using singular value decomposition has been applied to joint angles of the lower limb as representative kinematic parameters, joint moments related to the rotational motion of the joints have not yet been considered. In this paper, joint moments as kinetic parameters applied on the lower limb during gait of a normal subject and a trans-femoral amputee are analyzed under change in walking velocity by the wearable gait motion analysis system, and the effectiveness for quantitatively evaluate the rotational motion pattern in the joints of the lower limb by using joint moments is validated.

Measurement of rehabilitation in thumb MP joint subluxation due to rheumatoid arthritis.

As treatment for subluxation due to rheumatoid arthritis (RA), rehabilitation by hand therapy is one option, but the number of therapist is not sufficient. Therefore, a device for rehabilitation of thumb metacarpophalangeal (MP) joint subluxation has been developed. To improve the device, it is necessary to measure in close proximity to the actual rehabilitation. Therefore, the authors tried to measure two kinds of rehabilitation by using motion capture and a contact force sensor. To measure rehabilitation movements, three markers were attached to the metacarpal bone, six markers were attached to each side of the interphalangeal (IP) joint, MP joint and proximal phalanx of the right thumb of the subjects, and a finger model was created by these markers. Further, three markers were placed on the left index of the therapist, and force direction was calculated by these markers. Measurement was conducted on healthy subjects, Rehabilitation was performed by the person who is not a therapist, but received the guidance of the doctor who is coauthor. As a result, the authors could measure rehabilitation by hand therapy, force, point of action and displacement. The results suggest that rehabilitation with traction twice as efficient as that without traction. Furthermore, it was found that rehabilitation is possible with calculated force, and the force is reproducible by the actuator in the device.

Calculation of joint reaction force and joint moments using by wearable walking analysis system.

In gait analysis, which is one useful method for efficient physical rehabilitation, the ground reaction force, the center of pressure, and the body orientation data are measured during walking. In the past, these data were measured by a 3D motion analysis system consisting of high-speed cameras and force plates, which must be installed in the floor. However, a conventional 3D motion analysis system can measure the ground reaction force and the center of pressure just on force plates during a few steps. In addition, the subjects' stride lengths are limited because they have to walk on the center of the force plate. These problems can be resolved by converting conventional devices into wearable devices. We used a measuring device consisting of portable force plates and motion sensors. We developed a walking analysis system that calculates the ground reaction force, the center of pressure, and the body orientations and measured a walking subject to estimate this system. We simultaneously used a conventional 3D motion analysis system to compare with our development system and showed its validity for measurements of ground reaction force and the center of pressure. Moreover we calculated joint reactions and joint moment of each joint.

Gait motion analysis in the unrestrained condition of trans-femoral amputee with a prosthetic limb.

Trans-femoral amputees must regain moving pattern by refined rehabilitation program using ground reaction forces, joint angles and joint moments applied on a prosthetic limb. On the other hand, understanding those loads and kinematic variables is indispensable for gait analysis based on the biomechanical consideration of trans-femoral amputees. However, conventional prosthetic gait training systems cannot measure long continuous walking motions. In this paper, ground reaction forces and kinematic parameters applied on trans-femoral prosthesis are measured by the prosthetic gait motion analysis system using mobile force plate and attitude sensor for the unrestrained gait measurement. As a result of the experiments, the patterns of antero-posterior axis ground reaction forces and joint moments about the medio-lateral axis are remarkably different among the five activities. Finally, the effectiveness of the developed prosthetic gait training system to consider biomechanics and kinematics in trans-femoral prosthesis is validated.

Biomechanical consideration based on the unrestrained gait measurement of trans-femoral amputee with a prosthetic limb.

Trans-femoral amputees must regain moving pattern by refined rehabilitation program using loads applied on a prosthetic limb. On the other hand, understanding those loads is indispensable for biomechanical consideration of trans-femoral amputees. However, conventional prosthetic gait training systems cannot measure long continuous walking motions. In this paper, loads applied on trans-femoral prosthesis are measured by the prosthetic gait training system for the unrestrained gait measurement. As a result of the experiments, the patterns of moments about the medio-lateral axis are remarkably different among the six activities. Finally, the effectiveness of the developed prosthetic gait training system to analyze biomechanics in trans-femoral prosthesis is validated.

Development of a novel six-axis force/moment sensor attached to a prosthetic limb for the unrestrained gait measurement.

Since the number of trans-femoral amputees has increased by industrial or traffic accidents in modern society, a prosthetic limb has been required. In this case, those amputees must regain moving pattern by efficient gait training using load conditions on a prosthetic limb as quantitative evaluation indices. However, conventional gait training systems cannot measure long continuous walking motions. In this paper, a novel six-axis force/moment sensor, which is attached to a prosthetic limb for the unrestrained gait measurement, is developed. As a result of applying response surface method and desirability function, optimum design variables to reduce interference components are obtained. Finally, characteristics test by applying optimum design variables is performed and the effectiveness of the developed sensor is validated.

Development of walking analysis system consisting of mobile force plate and motion sensor.

In walking analysis, which is one useful method for efficient physical rehabilitation, the ground reaction force, the center of pressure, and the body orientation data are measured during walking. In the past, these data were measured by a 3D motion analysis system consisting of high-speed cameras and force plates, which must be installed in the floor. However, a conventional 3D motion analysis system can measure the ground reaction force and the center of pressure just on force plates during a few steps. In addition, the subjects' stride lengths are limited because they have to walk on the center of the force plate. These problems can be resolved by converting conventional devices into wearable devices. We used a measuring device consisting of portable force plates and motion sensors. We developed a walking analysis system that calculates the ground reaction force, the center of pressure, and the body orientations and measured a walking subject to estimate this system. We simultaneously used a conventional 3D motion analysis system to compare with our development system and showed its validity for measurements of ground reaction force and the center of pressure.

Forearm motion discrimination technique using real-time EMG signals.

The objective of this study is to develop a method of discriminating real-time motion from electromyogram (EMG) signals. We previously proposed a motion discrimination method. This method could discriminate five motions (hand opening, hand closing, hand chucking, wrist extension, and wrist flexion) at a rate of above 90 percent from four channel EMG signals in the forearm. The method prevents elbow motions from interfering with hand motion discrimination. However, discrimination processing time of this method is more than 300 ms, and the shortest delay time that is perceivable by the user is generally regarded to be roughly 300 ms. Furthermore, a robot hand has a mechanical delay time. Thus, the discrimination time should be less than 300 ms. Here, we propose a real-time motion discrimination method using a hyper-sphere model. In comparison with the old model, the hyper-sphere models can make more complex decision regions which can discriminate at the state of the motion. Furthermore, this model can learn EMG signals in real-time. We experimentally verified that the discrimination accuracies of this method were above 90 percent. Moreover, elbow motions did not interfere with the hand motion discrimination. The discrimination processing time was less than 300 ms, and was about 30 percent shorter than that of the old method.

Development of prosthetic arm with pneumatic prosthetic hand and tendon-driven wrist.

Recently, various prosthetic arms have been developed, but few are both attractive and functional. Considering human coexistence, prosthetic arms must be both safe and flexible. In this research, we developed a novel prosthetic arm with a five-fingered prosthetic hand using our original pneumatic actuators and a slender tendon-driven wrist using a wire drive and two small motors. Because the prosthetic hand's driving source is comprised of small pneumatic actuators, the prosthetic hand is safe when it makes contact with people; it can also operate flexibly. In addition, the arm has a tendon-driven wrist to expand its motion space and to perform many operations. First, we explain the pneumatic hand's drive mechanism and its tendon-driven wrist. Next, we identify the characteristics of the hand and the wrist and construct a control system for this arm and verify its control performance.

Prosthetic hand control using motion discrimination from EMG signals.

In this report, we improve the motion discrimination method from electromyogram (EMG) for a prosthetic hand and propose prosthetic hand control. In the past, we proved that a motion discrimination method using conic models could discriminate three hand motions without the incorrect discriminations that the elbow motions cause. In this research, to increase discrimination accuracy of motion discrimination using conic models, we propose a feature extraction method using quadratic polynomials. Additionally, because many prosthetic hands using motion discrimination have constant motion speed that can't be controlled, we propose an angular velocity generation method using multiple regression models. We verified these methods by controlling the 3D hand model. In the experiment, the proposed method could discriminate five motions at a rate of above 90 percent without the incorrect discriminations that elbow motions cause. Moreover, the wrist joint angle of the 3D hand model could be controlled by standard variation of 3[deg] or less.

Hand motion discrimination by EMG signals without incorrect discriminations that elbow motions cause.

The objective of this research is to develop a hand motion discrimination method that uses electromyogram (EMG) signals for controlling myoelectric prosthetic hands without the incorrect discriminations that elbow motions cause. We proposed a method that can discriminate hand motions using linear multiple regression models in the past. However, this method incorrectly discriminates hand motions when elbow motions are performed. The incorrect discriminations hamper myoelectric prosthetic hand use. To solve this problem, we propose a hand discrimination method using conic models. In the experiment, we prove that the proposed method can discriminate hand motions without the incorrect discriminations.