Robot Automatic Path Planning And Constant Force Control Based On Force / Visual Guidance

The traditional manual teaching and off-line programming methods are difficult to be applied to the robot task of small batch,multi variety and non-standard workpiece,so it is necessary to introduce external perception to construct a robot operation system based on intelligent guidance.This paper takes grinding process as an example to describe and analyze the system in detail.For workpieces with arbitrary curved shapes,on the one hand,the complexity of the curved surface shape makes the planning of the grinding path more difficult,on the other hand,it also has a certain impact on the force in the grinding process,therefore,the system not only needs to consider the adaptive path generation for curved workpiece,but also must consider the constant force control during robot grinding process.In addition,most current robot manufacturers only open the position control loop and do not open or provide the force control interface,which requires the system to achieve constant force control based on the robot position loop.In this context,the linear structured light sensor and force sensor are used as sensing tools to complete the robot automatic path planning and constant force control based on force/vision guidance.The robot grinding path generation based on surface adaptation and robot high-precision constant force control are realized.The details are as follows:Aiming at the problem of autonomous operation path generation of robot grinding system,this paper proposes a method for generating optimized trajectory of robot grinding based on surface reconstruction.First,scan the workpiece through the line structured light sensor to obtain the point cloud data of the workpiece surface;secondly,preprocess the original point cloud through point cloud simplification,point cloud interception,and point cloud filtering;then use the greedy projection triangulation algorithm to reconstruct the workpiece surface model;Finally,the truncated plane method based on the triangular mesh model is used to generate the reference trajectory for robot grinding,thereby realizing the generation of the grinding path based on the surface adaptation.In view of the characteristics that the robot end will be offset due to joint flexibility under the force,this paper proposes an offline constant force compensation control based on the robot joint stiffness.First,analyze and solve the robot kinematics model;then establish the robot joint stiffness model based on the relationship between the robot end position offset and the force applied to the end;finally,an offline constant force compensation control method based on the robot joint stiffness model is proposed,which can make the relative force error within 5.5% without force sensor,and meet the requirements of industrial production for grinding accuracy.In order to further improve the force control accuracy of the robot,this paper takes the robot as the control object and proposes three model-free online force tracking methods based on force sensor feedback.Firstly,the PID constant force control method is proposed according to the PID parameter model;then the traditional admittance control method is described and analyzed,on this basis,the variable stiffness adaptive admittance control is proposed,and the system stability proof is given by combining with Lyapunov stability theory,which can realize the rapid adjustment of contact force and effectively improve the jitter problem of force in the process of robot movement,eventually,the relative force error is controlled within 4.55%,which is improved compared with the offline constant force compensation control.On the basis of the above research content,this paper takes the existing laboratory ABB IRB120 robot and several workpieces as the research object,the reconstruction of the researched workpiece surface model,the generation of the optimized trajectory of the robot grinding,and the offline force compensation based on the robot joint stiffness Control and robot online constant force control have been experimentally studied.The experimental results verify the feasibility and effectiveness of the method proposed in this paper,indicating that this paper has some application value for the related research on robot grinding system.