Research On The Moving Intention Recognition Method For The User Of A Walking Robot

Due to the development of aging society,the demand for medical services to help the elderly has greatly increased.For many elderly people and people with walking disabilities,the reason for the decline in self-care ability is that they cannot walk alone.In the face of such problems,assisting robots provides a solution.However,an important research direction is how to improve the efficiency of the robot in assisting the user in daily life or in the course of rehabilitation training,so increasing the accuracy of the intention recognition of the user's movement direction is important.At present,in the study of human-computer interaction methods for assisting robots,there are methods of using a joystick,speech recognition,brain electricity,and myoelectric signals.However,there are problems such as inconvenience or high prices.In the study of this paper,the pressure sensor on the assisting robot and the portable smart phone are jointly used to obtain the upper body posture angle and forearm pressure data of the robot user,so as to identify the movement direction intention of the assisting robot user.This thesis first introduces normal human walking posture and elder's walking movement posture,and then compares and finds the role of the human upper body trunk in maintaining human walking stability,and explains the rationality of the upper body posture as a source of data to obtain users' sports intention.According to the quaternion attitude calculation algorithm,the acceleration and angular velocity data obtained from the mobile phone are converted into the posture angle of the user's upper body.A kinematics model of the walking robot was established.Through experimentation,the front,back,left,right,left front,left rear,right front,and right sides of multiple experimenters using the walking robot were collected.The data of the movements of the back,left and right rotations were studied.The relationship between the experimenter's upper body posture angle and forearm pressure and the experimenter's movement direction intention was studied.The difference between the rotation movement intention and the omni-directional movement intention was analyzed.The support vector machine was used to Experimental omnidirectional motion intent data and left and right rotational motion intent data were classified.Because of the actual situation,the omnidirectional motion intention of the walking robot user is an arbitrary direction angle,that is,the direction angle belongs to any value between 0°and 360°.Therefore,after distinguishing the rotational motion intention from the omni-directional motion intention,a fuzzy rule base is established based on the relationship between the experimenter's upper body attitude angle and forearm pressure and omnidirectional intention,and a distance-based fuzzy inference algorithm is used to achieve The experimenter's movement data at one moment is the same as the experimenter's omnidirectional movement intention.Finally,experiments verify the feasibility and effectiveness of the proposed algorithm.