Research On Human-Machine Cooperative Control System For Excretory Support Robot

Against the background of an elderly population,a shortage of social resources and artificial intelligence pension,to alleviate the burden of caring for the elderly and the disabled,improve work efficiency,the development of cooperative robots that help caregivers reduce the burden of toilet care is of great significance for improving the quality of life of the elderly and the disabled and improving the utilization of social resources.Based on the consideration of appropriate location,rapidity and safety,the human-machine cooperation process is modeled and analyzed.Firstly,the hardware and software platforms for the experiment are built on the basis of the excretory support robot independently developed by the research group.The robot adopts omnidirectional wheel structure,which has omnidirectional motion function and can move flexibly in a narrow space.In order to enable the robot to better complete the human-machine cooperation task,the relative position of the human and the robot is sensed by two single-line laser radars installed in front of the robot.Using the laser radar below,the position of the human leg is obtained by object segmentation and circle-like detection techniques.The mode distribution of leg position in twoperson cooperation is presented,the relative position information between the user and the robot can be detected accurately by combining the detection information of the above lidar.In the case of Human-robot cooperation and normal sitting down,the motion characteristics of human body during sitting down are analyzed and an improvement on the original two-link model of human body is proposed.The mapping table of human body position and sitting point is established by the human body sitting model and the physical size of the robot.Based on the idea of reinforcement learning,according to the motion of the robot,the reward for reinforcement learning is given by fusing subjective and objective factors with fuzzy reasoning,and the strategy for finding the appropriate sitting point is formulated.Thus,the mapping table can be adjusted by self-learning.The kinematics model of foot trajectory of user in the process of transfer is analyzed and established,which can be used to predict the foot trajectory and improve the response speed of robot.Aiming at the path planning problem of robot in motion,this thesis divides the motion planning into rough planning and fine adjustment.The rough planning is to analyze the position relationship when of the robot position exceeds the sitting position of the human body sitting model and path planning using virtual target points and artificial potential field.The fine adjustment refers to the fine adjustment of robot position by the sitting point calculated from the human body sitting model.In view of safety,a method for calculating the variable safety distance considering the robot speed and obstacles is proposed to ensure the safety of the robot in human-machine cooperation.The data processing,model building and simulation of each chapter in this thesis are all carried out under the environment of Matlab 2017.vs2013 is the software platform and C++ is the programming language for the actual control and operation of robots.The experimental results show that the robot can detect the location information of human and move to the right position safely and quickly.Experiments verify the rationality and feasibility of the proposed model and strategy.