Research On Active Compliance Control Of Large Deformation Parallel Mechanism

As collaborative robots assisting human beings in areas such as industry,medical treatment and home care,the safety of human-machine collaboration needs higher requirements.Compliance mechanism is an effective method to improve the safety of man-machine collaboration,which mainly includes passive compliance and active compliance.Passive compliance complies with contact force by means of internal elastic elements.The control mode is simple and fast,but its application is limited and stiffness is difficult to adjust.Active compliance is also widely used in industry.The response bandwidth is limited due to the lack of elastic elements.Its adaptability depends heavily on the reliability of algorithms and sensors.In order to solve the compliant mechanism problems,a flexible linkage mechanism with adjustable stiffness and wide range of motion is designed,which has the ability of force sensing.This is of great significance to the development of cooperative robots.Therefore,a three-degree-of-freedom flexible link parallel mechanism with large deformation and force sensing capability is proposed.It uses the coordinated deformation of multiple elastic plates to make the mechanism end move in a large space.After the analysis of elastic materials,cross section and mechanism configuration,a redundant driving prototype is designed and manufactured.The main research contents are as follows:(1)Theoretical model of the deformation and movement of the parallel mechanism with flexible linksUsing principal axes decomposition of compliance matrices,the elastic rod is divided into several passive series joints.This section studies the kinematic of mechanism with single plate,double parallel plates and four parallel plates.The multi-objective optimization model is established by deriving the configuration constraint equation and static balance equation of the mechanism,while the solution is efficiently solved via the gradient equation.The experimental platform is designed and established.The accuracy of repeated and absolute positioning is 0.707 mm and 1.53 mm respectively,which proves the accuracy and reliability of theory model.The adjustable stiffness of the redundant driving device is verified by mechanical experiments.(1)Study on fast force sensing model of ”attitude-load” mapping based vision methodAccording to the characteristics of rigid flexible coupling mechanism,the load force can be estimated indirectly by observing the pose of rigid links.Firstly,the pose measurement function based on feature points is realized.Compared with the API laser tracker,the error is within 0.6mm.Then,the vision sensor is arranged on the parallel mechanism.Compared with the instrument,the positioning error of vision sensor is less than 1.5mm.In order to avoid complex problems such as low calculation efficiency,random mutation of contact force and machining error,multi-layer neural network is applied to fit the ”attitude-load” mapping,which improves the calculation efficiency and robustness of force sensing.The theoretical data set of mixed noise is transmitted to the network for supervised learning.After training,the error of network is less than 0.1,the experiment error is less than 10%,and the response speed is 10 ms,which ensures the accuracy and rapidity of force sensing.(3)System development and experiment research of flexible link parallel mechanismAccording to the principle of modularity and loose coupling,the operation software is designed,and the human-computer interface is compiled,which is convenient for human operation and data monitoring.The experiments of shaft-hole assembly and glass objects operation are successfully carried out,which verify that the designed flexible linkage mechanism has excellent movement ability and compliance.