Font Size: a A A

Deformation Tracking And Manipulation Of Unmarked Deformable Planar Object

Posted on:2020-10-10Degree:MasterType:Thesis
Country:ChinaCandidate:M Y LiFull Text:PDF
GTID:2428330590973955Subject:Mechanical and electrical engineering
Abstract/Summary:PDF Full Text Request
An object whose deformation must be considered during manipulation is referred to as a deformable object.According to this,it can be divided into a linear deformable object,a planar deformable object,and a solid deformable object.The planar deformable object is widely used in life,and the fabric,the paper sheet,the flexible circuit board and the like are all planar deformable objects.At the same time,with the development of robot technology,robot technology not only has a wide range of applications in the classic application fields of welding,painting and handling,but also plays an important role in additional fields such as assembly,processing and measurement.There are many recent research or projects aimed at exploring innovative applications in the robotics arena.The robotic manipulation of the planar deformable object is an innovative application in the field of robotics.This innovative application has great potential for application in domestic service and industrial production.Based on the above background,a method for tracking the unmarked planar deformable object is proposed,and the shape control of the planar deformable object is realized by controlling the mechanical arm.The tracking of the deformable object requires a non-deformed state as a reference,and the process of deformation is also memory.The traditional deformation tracking is to identify the feature points on the deformable object.These feature points can still be traced before and after the deformation.The changes of the feature points before and after the deformation reflect the changes of the deformable object.This paper proposes a NURBS surface fitting to track the deformation of the unmarked planar deformable object.The difficulty is that there are no obvious feature points for estimating the deformation state of the deformable object.In this paper,the unmarked planar deformable object tracking is divided into two parts,the static deformable object deformation tracking and the deformable object deformation tracking in motion.The stationary deformation tracking method is a dynamic NURBS surface fitting method which tracking the deformation of a point cloud of a stationary planar deformable object.The method ensures that the deformation of the fitted surface is consistent with the actual deformation of deformable object,and the deformation of the fitted deformable sur-face can be obtained by calculating the deformation of the fitted surface.The deformation tracking method in motion uses the relationship between the robot arm and the deformable object to decouple the deformation information and the movement information in the point cloud,and solves the problem of the movement of the labelless surface deformable object.When the control arm manipulates the planar deformable object,a deformed Jacobian matrix is used,and the correspondence between the end speed of the arm and the key point speed on the planar deformable object is established.In combination with the differential kinematics of the manipulator,the velocity of a point on the planar deformable object is mapped to the angular velocity of the joint space of the robot arm.A controller is designed to congtrol the robotic arm which manipulates the planar deformable object.In order to verify the effectiveness of the method described in this paper,a visual servo experimental platform based on the robot operating system(ROS)was built.The visual servo labeling experiment of unmarked paper was carried out on the platform,and the experimental results were analyzed.The experimental results reflect that the method described herein is capable of tracking the deformation of the paper well and performing the desired deformation of the planar deformable object by the visual servo.
Keywords/Search Tags:deformable planar object, point cloud processing, NURBS, deformable object manipulation
PDF Full Text Request
Related items