Study On Eye-in-hand Vision Guiding Technique Of Baxter Dual-arm Robot For Grasping And Peg-in-hole Assembly

Due to its advantages of high flexibility and coordination, Redundant dual-arm robot is widely applied in industrial automation, humanoid service, space exploration and medical equipment etc., playing an important role in the rising robotic research field. It’s of significance to study an integrate and simple method for part grabing and assmbling, since it’s the common and crucial tasks of robot application.The research of this paper is based on Baxter robot with dual arms of 7 vertical joints, it’s software system is built based on ROS(Robotic Operating System). Firstly, analyzes the kinematic of Baxter according to URDF(Uniform Robot Description Format), using gradient projection method to get inverse-kinematic solution of optimization in velocity level, the solution is then put forword by Newton-Raphson iteration. The calibration model is built after adding geometrical constraint in kinematic model, then using screw-axis measurement method and kalman filtering method to make kinematic calibration. Secondly, does part grabing using eye-in-hand system with dual arms and single arm respectly, then making comparation of these two methods. In order to improve the positioning accuracy, proposes a secondary positioning method. Finally, robot grabs parts to do peg-in-hole assembly with dual arms. The pose of right gripper relative to left gripper is measured using PnP method, transforming to compensation of right arm’s joints through differential inverse kinematics. Due to the flexibilty of dual arms, Baxter is able to accomplish more precise assembly task.Simulation and experiment are made by Matlab software and Laser Tracker, confirming that the algorithm of twice calculation to solve inverse-kinematic and the calibration method combining screw-axis measurement and kalman filtering are feasible. Cylinder part grabing and peg-in-hole assembly are also done, conclude that binocular hand-in-eye vision has lower position precision than monocular vision but performs higher roughness in complex environment, and secondary position can improve grasping precision effectly. The dual-arm assembly method based on eye-in-hand vision can correct the error efficiently and complete high precision assembly task. It’s applicative widely for dual-arm assembly.