Interaction Method Of Multiple Welfare-Robots Oriented To Safe Transfer Tasks

Under the background of aging society,rehabilitation robots and assisted robots have become a research hotspot and focus in the field of robotics.However,there are still some deficiencies in scene function and security.To fulfill the requests of the elderly and the disabled people with weak motion capability,the research and development of welfare robot and compliance control method are crucial to solve the issue of assisting the elderly and disabled in smart house,pension center,hospital and other places.In this dissertation,five welfare robots are developed and taken as the research objects.According to the characteristics of robot structures,kinematics and dynamics modeling,transfer motion planning,human-robot interaction and intelligent control are deeply studied,which contributes to lay the foundation for the application of welfare robots and safe auxiliary behaviors in daily life.The main research of this dissertation includes the following aspects:(1)To make up for the lack of scene function of traditional rehabilitation robots,aiming at the safe transfer of people with mobility difficulties in getting up,standing and excreting in daily life,smart house for the elderly and the disabled is established based on five welfare-robots and distributed sensor network.The whole system provides service platform for the transfer behaviors.It is a novel solution paradigm for behavior assistance in the field of assisting the elderly and disabled.(2)To achieve the safe rising-standing transfer,the human-robot integrated position model of rising-standing transfer and the motion planning of the system are established,which lays the foundation for the research on control strategy of transfer.A three-dimensional two-link seating model with position compensation is applied to calculate accurate transfer point.The contour recognition algorithm of lower limb based on object segmentation and circle-like detection is proposed.The two-level joint probabilistic data association algorithm with multi-constraint contour is proposed,which provides a highly robust solution for dynamic pose recognition under the condition of limb confusion,trunk occlusion and objects interference.The proposed method is applied to the rising-seating transfer.A follow-up control method of auxiliary excretory robot based on the dynamic pose of transfer is proposed,and the correctness and feasibility of the method are proved by transfer experiments.(3)A novel non-contact standing-walking transfer method is proposed for the compliant transfer behavior.This method ensures the safe transfer,and controls the robot to walk in all directions smoothly based on the user's intention.The fuzzy system based on the rule evolution strategy for directional intention recognition of walking is proposed to accurately identify the intention of walking direction and solve the sparse characteristic of data rules in this computational scenario.To reduce the reasoning errors caused by different interaction habits and unstable fuzzy rules,the rule evolution algorithm and knowledge radius algorithm are proposed to optimize the fuzzy rule base adaptively.The transfer experimental results show the proposed algorithm can effectively improve the adaptability and accuracy of dynamic intention recognition.(4)For safe standing-walking transfer,a non-contact abnormal behavior recognition method based on node-iterative fuzzy Petri net is proposed.To effectively extract the dynamic characteristics of various abnormal behaviors,the coordinated state eigenvalues of arm-trunkgait are established based on the deviation parameters of walking intention,body incline angle and dynamic frequency.The node iteration strategy is applied to the fuzzy Petri net algorithm,and create network ignition mechanism with adaptive optimization nodes.Experimental results show that the proposed method effectively improve the recognition accuracy and adaptability of the drag-to-drop gait and other abnormal behaviors,and improves transfer safety.Meanwhile,the non-contact recognition method has universality and convenience,and can be applied to all similar structures of robots.(5)To verify the safety,feasibility and effectiveness of the proposed method,the experiment of complex behavior integrated multi-transfer tasks based on multiple welfare-robots system is implemented.Based on the internal and external factors of human body,this dissertation comprehensively evaluates the safety of transfer.Through the comparative analysis of the pressure signal,myoelectric signal and cerebral blood oxygen signal of different difficulty levels of transfer tasks,it is verified that the proposed method can improve the safety of the users.Meanwhile,the proposed flexible control method of multiple welfare-robots oriented to multi-transfer tasks has a good promotion for hospitals,families and other disabled fields.