This paper presents an extended Kalman filter for pose estimation using noise covariance matrices based on sensor output. Compact and lightweight nine-axis motion sensors are used for motion analysis in widely various fields such as medical welfare and sports. A nine-axis motion sensor includes a three-axis gyroscope, a three-axis accelerometer, and a three-axis magnetometer. Information obtained from the three sensors is useful for estimating joint angles using the Kalman filter. The extended Kalman filter is used widely for state estimation because it can estimate the status with a small computational load. However, determining the process and observation noise covariance matrices in the extended Kalman filter is complicated. The noise covariance matrices in the extended Kalman filter were found for this study based on the sensor output. Postural change appears in the gyroscope output because the rotational motion of the joints produces human movement. Therefore, the process noise covariance matrix was determined based on the gyroscope output. An observation noise covariance matrix was determined based on the accelerometer and magnetometer output because the two sensors’ outputs were used as observation values. During a laboratory experiment, the lower limb joint angles of three participants were measured using an optical 3D motion analysis system and nine-axis motion sensors while participants were walking. The lower limb joint angles estimated using the extended Kalman filter with noise covariance matrices based on sensor output were generally consistent with results obtained from the optical 3D motion analysis system. Furthermore, the lower limb joint angles were measured using nine-axis motion sensors while participants were running in place for about 100 s. The experiment results demonstrated the effectiveness of the proposed method for human pose estimation.

This paper proposes a method of estimating the knee joint angle during walking using nine-axis motion sensors in a varying magnetic field. The nine-axis motion sensor comprises a three-axis gyro sensor, a three-axis acceleration sensor, and a three-axis geomagnetic sensor. It can estimate joint angles during exercise by correcting the drift of the three-axis gyro sensor using information obtained from the other two sensors. However, the magnetic field cannot be measured correctly using a three-axis geomagnetic sensor in a variable magnetic field. Therefore, the joint angle estimation accuracy is lowered. For this study, the authors corrected the outputs of a three-axis geomagnetic sensor using the three-axis angular velocity obtained from a three-axis gyro sensor. During the laboratory experiment, the 3D motion analysis system and two nine-axis motion sensors measured walking exercise. The knee joint angle results estimated using the two nine-axis motion sensors using corrected outputs of a three-axis geomagnetic sensor generally agreed with the 3D motion analysis system results. Furthermore, two nine-axis motion sensors measured walking exercise outside for about one hour. In the results of knee joint angle estimation using uncorrected outputs of a three-axis geomagnetic sensor, the boundaries between swing phases and stance phases were unclear. Results of knee joint angle estimation using corrected outputs of a three-axis geomagnetic sensor indicate a similar tendency to that found for results of the walking cycle from the laboratory experiment that comprised swing phases and stance phases. This analytical method is anticipated for use in estimating motion in a varying magnetic field.

This paper describes the use of nine-axis motion sensors to evaluate knee joint angle estimation accuracy during walking. The nine-axis motion sensor comprises a three-axis gyro sensor, a three-axis acceleration sensor and a three-axis geomagnetic sensor. It can estimate joint angles during exercise by correcting the drift of the three-axis gyro sensor using information obtained from the other two sensors. Human movement results from the rotational motion of the respective joints, so that the proportion of the centrifugal acceleration and the tangential acceleration in the output of the acceleration sensor increases during exercise. Processing the centrifugal acceleration and tangential acceleration appropriately and ascertaining the degree of estimation error are important for improving the joint angle estimation accuracy. For this study, the authors produced a sensor fusion algorithm using an extended Kalman filter to correct the acceleration sensor output. The sensor fusion algorithm uses information obtained from the nine-axis motion sensors to estimate the knee joint angle by correcting the centrifugal acceleration and tangential acceleration. During the experiment, the 3D motion analysis system and two nine-axis motion sensors measured walking exercise. The knee joint angle was estimated using an extended Kalman filter with information obtained from the nine-axis motion sensors. We evaluated the system accuracy for knee joint angle estimation by comparing the nine-axis motion sensor results and the 3D motion analysis system results. This analytical method is anticipated for use in estimating motion in sports and healthcare applications.

ADW is a fixed abrasive diamond saw wire manufactured by brazing diamond grains to a metal wire. Brazing makes it possible to firmly bond a diamond grain onto a metal wire. Therefore, ADW has a longer operating life compared with conventional electroplated diamond wire saws. However, it was found that when this tool is used on a single track with larger cutting force, there is the possibility of the tool breaking easily from the past experiments. Therefore, we developed twisted ADWs for improving the strength of ADW. In this study, good cutting conditions, such as cutting force, tension, wire speed, were investigated experimentally for high speed cutting performance with twisted ADWs. Observing the SEM photographs of used twisted ADWs for cutting, the contact state of diamond grains and brazing metal, the behavior of diamond grains in cutting were considered. As the result, the best cutting conditions (cutting force is 30 N, tension is 30 N, wire speed is 150 m/min) were found from two viewpoint of cutting performance and operating life.

This paper presents a proposed method for center of gravity (COG) velocity estimation during squatting using information obtained from lower limb motion measurements. A squat exercise uses the lower limb joints and the muscles around these joints. The squat exercise velocity changes according to the lower limb muscle activity. Some earlier reports of relevant studies have suggested that muscle weakness and neurological deterioration influence the COG velocity when standing up. Therefore, it is important to clarify the relation between the COG velocity and lower limb motion for efficient training and the prevention of falls among older people. For this study, we constructed a squat velocity model that represents the relation between the COG velocity and lower limb joint power during squatting. Although no joint power indicates each muscle activity in detail, it is possible to estimate muscle activities around the joints approximately by using the joint power. For this experiment, the squat exercise was measured using a 3D motion analysis system. The experiment was conducted for three stance widths. The COG velocity and the lower limb joint powers were calculated using information from the 3D motion analysis system. We estimated the squat velocity model parameters by application of Kalman filter using the measurement information. The analysis results for the squat velocity model indicated a quantitative relation between the COG velocity and the lower limb joint power during squatting. Furthermore, comparable results were obtained from three stance widths. This analytical method is anticipated for use in evaluation of motor function and exercise assisting device design.