Research On Key Technologies Of Robot Grabbing In Nuclear Decommissioning Scenarios

With the rapid development of the nuclear industry,there exists many nuclear facilities in China that need to be decommissioned urgently.Nuclear decommissioning robots can replace manual labor and complete decommissioning work in a highly radioactive environment.However,the traditional teleoperation control method has problems such as difficult operation,low efficiency,and long personnel training period,so the autonomous operation in nuclear environment is one of the research focuses of nuclear robots.This paper aims to improve the autonomous grasping operation capability of nuclear robots,focusing on the path planning problem,trajectory tracking control problem and grasping position detection problem faced by robots in the process of autonomous operation.The main contents of this paper are as follows:(1)The nuclear robot kinematics and grasping problems are analyzed for the specificity of nuclear decommissioning scenarios.Firstly,the grasping problem faced by the robot in the nuclear decommissioning scenario is analyzed.Then,an omnidirectional robot is selected as the object of study and its kinematic model is established to be applicable to the narrow operation space.(2)To address the problems of slow planning,random node expansion,high path cost and smoothness of the Rapidly Exploring Random Tree Star(RRT*)algorithm,a bidirectional RRT*(Target gravitational)algorithm based on improved metric function and target gravity is proposed.Target Bidirectional RRT*(TB-RRT*)algorithm based on improved metric function and target gravitational force is proposed.By using the target gravitational force and dynamic step method,the algorithm can improve the operation efficiency and reduce the path cost.The smoothness of the path is improved by improving the metric function to consider the effects of both Euclidean distance and pinch angle on the path planning,and the planned path is transformed into an executable trajectory of the robot by the fifth polynomial time scale method on this basis.Finally,the effectiveness of the algorithm is verified by simulation experiments.(3)In order to improve the robot operation efficiency,a trajectory tracking controller and a grasping strategy are designed.For the problem of difficult and time-consuming position adjustment caused by the narrow operating space of the robot,a cooperative control strategy between the robot motion platform and the robotic arm is designed to achieve accurate control of the end-effector through the combination of motion between the motion platform and the robotic arm to improve the operating efficiency.Meanwhile,considering the strong nonlinearity of the robot system,a nonlinear PID trajectory tracking controller based on feedforward-feedback is designed.To obtain a more reasonable grasping posture when performing the grasping task,a grasping posture detection method based on Res Net34 is used.Finally,the effectiveness of the controller is verified by simulation experiments.(4)The feasibility of this paper’s method is verified by conducting visual simulation experiments through Coppelia Sim.The experimental results of path planning show that the TB-RRT* algorithm proposed in this paper has 7% shorter path length,40% fewer sampling nodes and 40% less computing time compared with the(Bidirectional RRT*,B-RRT*)algorithm.The experimental results of trajectory tracking show that the designed controller has smaller error.The experimental results of the grasping pose detection show that it can effectively achieve the grasping pose detection of the target object,and the detection success rate is 81.94%.The experimental results of grasping show that the designed grasping strategy can achieve a grasping success rate of 76.25%.