Research On Assembly Method Of Shaft Parts Based On Vision And Force Sense

The development of robots has empowered the manufacturing industry to upgrade and upgrade in leaps and bounds,but with the complexity of application scenarios,traditional robots controlled solely by position control can no longer meet the demand.Combining robots with vision and force sensing plays an important role in improving robot flexibility,intelligence,and reliability,and is now one of the popular research areas.Robots with vision and haptics have the ability to complete tasks in uncertain environments.The assembly of shaft parts is a common application of robots in industrial production,and proposing an effective intelligent assembly method for shaft parts can greatly improve the efficiency in industry,which is of great importance for intelligent production.In this paper,we explore an automatic assembly algorithm based on vision and force perception for shaft parts.We combine a six-axis industrial robot with a depth camera and a six-dimensional force sensor to build an assembly system that includes three subsystems:a vision perception system,a force perception system,and a robot control system.The main tasks accomplished in the thesis are as follows.1)For arbitrarily placed part poses,this paper constructs a vision perception system based on Ensenso N10 depth camera to achieve object detection and positioning.The camera is mounted in an eye-in-hand manner,and the position of the camera in the robot end coordinate system is firstly obtained by hand-eye calibration.Using the results of hand-eye calibration and robot posture,the point cloud data obtained by the depth camera can be directly converted into the robot base coordinate system.2)For the error of camera shooting and complex scenes,this paper discusses the point cloud segmentation and filtering algorithm.In the segmentation stage,the combination of supervoxel segmentation together with locally connected convex patches is selected through analysis and comparison,and the required data parts are successfully separated.In the filtering stage,the noise components of the point cloud are analyzed,and the combination of supervoxel fitting plane and non-maximum suppression algorithm to filter the plane is proposed to obtain a smooth and noise-free part point cloud.3)In order to obtain the accurate poses of the parts,the point cloud alignment algorithms are analyzed and compared in this paper.In the rough alignment stage,two local feature-based alignment methods and the principal component analysis method are introduced and experimentally compared.For accurate alignment,this paper analyzes in detail the principle,advantages and disadvantages,and improvement points of the iterative nearest point algorithm,and proposes a new error function applied to the original algorithm,which solves the ineffectiveness of the algorithm for point cloud data with low overlap rate.Experiments show that the combination of the principal component analysis method and the improved iterative nearest point algorithm can achieve the required positional accuracy for part assembly.4)In order to realize the robot's perception of external forces,a force perception system based on six-dimensional force sensors is constructed in this paper.Firstly,in order to accurately perceive external forces,it is necessary to remove the system errors and the influence of gravity of end tools and parts,so the algorithm of zero calibration and gravity compensation of force sensor is proposed,during which the angle between robot and horizontal plane is fully considered,and finally the robot can obtain accurate information of contact force of parts.5)For the robot flexible assembly task,this paper proposes an active flexible control method for the robot based on the admittance principle.By inputting the deviation of the desired force from the actual force into the system,the desired position of the robot is adjusted and the moving target of the robot is adapted in real time,which is verified by force control experiments that the robot has the characteristic of moving with the force,that is,it exhibits the desired flexible characteristic.