Research On Measurement Of Foot Posture And Ground Reaction Force For Clinical Gait Analysis

This paper developed a pair of wearable sensor shoes,which combined the function of motion capture system and force plates in the sensors shoes.The purpose of this research is real-time successive measurement of foot's orientation and ground reaction forces in daily-life enviroments.The sensor shoes',soles are capable of measurement of 3 axial ground reaction forces and moments,inertial sensors are mounted inside shoes for measurement of orientations.Based on the development of this sensor shoes,the main contents of this paper include:1.A fusion method is proposed to use both inertial sensor and range sensors for measuring orientation of foot.Compared with the results of 9 axial fusion methd,the accuracy of the attitude angles is improved.The commonly used 9 axial fusion method is susceptible to the interference of motion acceleration and external magnetic field.The new method is based on the extended Kalman filter algorithm and fuses the measurements of the range sensors and the angular velocity of the gyroscope for better solution of orientation.The results show that the fusion method can improve the accuracy of the pitch angle and roll angle effectively.2.An optimization method for anatomical calibration of foot joint is implemented,the main purpose of this procedure is to calculate the relative orientation between the segment's coordinate and the inertial sensor's coordinate.Compared with the manual calibaration and the calibration by predefined motions,this method requires no complicated procedures and the results are more reliable.The optimization method is based on the mathematical relationship between the joint coordinate system and the measurements,the gradient descent method is used to calculate final results.3.An inverse dynamics method is proposed for wearable sensor systems,the resultant ground reaction force is obtained by measuring the heel,arch,and forefoot respectively.Based on the ground reaction forces measured by the sensor shoes and the lower limbs,joint angles measured by the inertial sensors,the joint forces and joint torques of the lower limbs are obtained by the inverse dynamics analysis.4.A method is proposed for the lower limb kinematics analysis based on the ground reaction forces measured by sensor shoes.The neural netword and convolutional neural network are used to fit the hidden relationship between the ground reaction force and the lower limb's joint angle.The neural network has achieved good test results in the prediction of individual models,but the generalization of its general model is lack of ability.Further,the convolutional neural network is used to train the general model,and the prediction results of the joint angles of unknown person is acceptable.The prediction accuracy of knee joint angle and hip joint angle for unknown object is improved to the correlation coefficients of 0.94,0.93 and the root mean square error of 7.89°,3.5 1° respectively.In summary,this paper focuses on the research of measurement of foot posture and multi-axial ground reaction force.The contents including the fusion method for foot posture,the anatomical calibration method of foot joint,the measurement of multi-axial ground reaction force and its application for lower limb's inverse dynamics,and the lower limb's kinematics by ground reaction force.The main contributions of this dissertation are as follows:an extended Kalman filter algorithm is proposed to correct the attitude of inertial sensor by fusing multiple range sensors;a method is proposed to measure the ground reaction force by dividing the foot into three parts,and the lower limb's inverse dynamics is implemented by using these measurements;neural networks and a convolution neural network are proposed to predict the lower limb's joint angles only by ground reaction force.