Research On Gait Recognition System Based On Plantar Pressure Information

Exoskeleton robot is the hotspot in the field of robot research,the key technology is gait recognition.At present,the design of information acquisition and recognition algorithm of gait recognition mainly focuses on foot pressure,inertial sensor and so on.Because of its strong anti-interference and fast corresponding speed,the pressure sensor has attracted more and more attention.At the same time,with the development of AI,machine learning algorithm has also been favored in gait recognition.However,the problem of parameter selection in machine learning also arises.In this paper,two key issues are studied,and a gait recognition system based on plantar pressure is designed.First of all,aiming at sensor placement,the foot and ankle bones of human lower limbs are anatomically analyzed,and the key supporting points of foot are obtained.Starting from ankle joint and based on the d'Alembert-Lagrange method,the dynamic model of the foot ankle system and seven links of human body are established.The relationship between the toe friction and the toe pressure of the single foot support phase and the gait phase is analyzed,which provides the theoretical basis for the placement of the foot pressure sensor.Secondly,aiming at the limitation of traditional gait recognition system,a gait recognition system based on plantar pressure is designed.The system consists of two parts:a gait data acquisition system and a gait recognition algorithm.The gait data acquisition system collects plantar pressure information and foot posture in the process of human walking.In this paper,STM32 MCU is used as the core control unit of the system,and gait data is transmitted through wireless module.In view of the instability of the original data,the data were cut,interpolated and segmented,and the human walking phase database was established.Finally,aiming at the problem of SVM parameter selection,an improved particle swarm optimization algorithm(PSO-GA)is studied and designed,on which the key parameters of SVM are based and optimized.Then gait recognition algorithm are completed.Finally,the performance experiment of gait phase recognition system is carried out,and the performance of the model is evaluated by two parameters of inter class recognition accuracy and average recognition accuracy.By comparing the performance of the gait recognition system based on PSO-GA-SVM and the gait recognition system based on PSOSVM,it is concluded that the designed system has better performance.