Design And Development Of A Robot Loading System For Bolt Hardness Testing Based On Visual Positioning

The hardness of military bolt fasteners is an important indicator of the safety performance of military aircraft.Manual batch testing of the hardness of the bolts is easy to cause fatigue to the operator,and there are problems of high labor cost and low efficiency.In response to the above situation,this paper carried out the design and development of a robot loading system for bolt hardness testing based on visual positioning.The main research contents are as follows:First,this article discusses and explains the specific requirements of the bolt hardness testing and feeding system,and designs the overall plan of the system.At the same time,the functional parameters of the equipment required by the computing visual recognition subsystem and the robot subsystem are checked separately,and the hardware platform is built.Furthermore,based on the interaction process between the subsystems,this paper designs the control system architecture and develops the upper computer software platform of the vision system.Secondly,the identification algorithm research is carried out for the presence of foreign objects and underfilled bolts on the pallet.Firstly,the two-dimensional code mark on the tray is used for positioning,and the geometrically distorted twodimensional code image is preprocessed and corrected based on affine transformation.Based on the principle of the smallest circumscribed circle,the operator is prompted to clean up after identifying foreign objects on the tray.When identifying whether the bolt is full,analyze and compare Hough circle detection and threshold-based recognition algorithms.The latter does not require strict morphological operations and has higher accuracy.Next,this paper aims at the problem of obstacle avoidance when there are obstacles in the robot movement process,and proposes a variable step size improved RRT* algorithm based on the idea of target bias.Perform D-H modeling on Estun ER12-1510 robot,analyze its kinematics,and use numerical methods to solve its workspace.The collision detection model is simplified into a collision method of a straight line in space and a sphere.The improved RRT* algorithm proposed in this paper is compared with the rapid expansion random tree(RRT)algorithm and RRT*algorithm.The results show that the improved RRT* algorithm has fewer overall sampling times and reduces the probability of invalid nodes.The planned path length is compared with RRT,RRT* algorithm reduces 28.28% and 17.72% respectively.Finally,a vision-based bolt hardness testing robot feeding system experiment platform was built to conduct visual positioning,bolt automatic feeding and robot obstacle avoidance experiments.The results show that the system can accurately identify and grasp bolts,and the robot can accurately avoid obstacles,verifying the effectiveness of the improved RRT* algorithm in this paper.It takes 14.3 s for the robot to complete a loading process with the longest motion path.The axial positioning accuracy of the bolt placement point is-0.102 mm ～ 0.044 mm,which meets the performance requirements.The successful construction of this system is of great significance in improving the efficiency of bolt hardness testing and reducing labor costs.