Research On Path Planning Of Mobile Manipulator Under Restricted Conditions

As China’s manufacturing is moving toward a more sophisticated and smarter direction,the robotics field has become a hot research area in China.As a combination of a mobile robot and a robot arm,the mobile manipulator is a robot that can autonomously sense,plan,and complete various tasks.For factories,logistics,and service industries that require flexible and autonomous robots to perform complex tasks,it is necessary and meaningful to research mobile manipulator.When the mobile robot arm is in operation,an optimal pre-operational posture is critical for accomplishing tasks such as grabbing.The traditional method is to adjust the posture to achieve the best when the mobile manipulator reaches the mission site.Position state,but when the task point space is small,the process of adjusting the posture will be very difficult and time consuming.If this process is not required,that is,when the mobile robot arm arrives at the task point from a non-task point or other task point,it is in the best task posture,which will greatly improve the efficiency of completing the task.This evolves into a problem of how to plan a path from one pose to another in a defined workspace.In this paper,the pose is used as a qualification condition.Taking the grabbing task of the mobile manipulator in the smart factory as an example,the path planning problem of the mobile manipulator under restricted conditions is studied.In this paper,the path planning problem under restricted conditions is decomposed into three parts: how to obtain environmental information and map automatically,how to get the best pre-crawling pose,how to plan a path in the map so that the mobile robot arm is in the pre-grabbing posture state from the initial pose to the end of the path.For these three problems,we first propose a detection strategy based on the improved Rapidly-exploring Random Trees(RRT)algorithm,which uses a global detection tree with variable growth rate to accelerate tree growth and The detection of the boundary points,at the same time,generates a local detection tree that follows the movement of the robot,prompts the robot to detect nearby unknown areas,and then uses the SLAM package in ROS to simultaneously establish a map,so that it can independently complete the exploration and mapping of environmental information.In a smart factory,the position of the workpiece and the task point is mostly fixed,and the spatial posture of the end effector of the moving robot arm is predictable when performing a specific task.According to this pose information and map information,we adopt the inverse kinematics solution method based on redundant parameters.By introducing redundant parameters and ensuring the convergence solution of the pre-crawling poses of redundant parameters can be obtained according to the geometric method.Redesign the optimization objective function to get the best pre-crawling pose solution.Then we introduce the artificial potential field function into the RRT-Connect algorithm,and propose the path from the initial pose to the pre-grabbing pose by proposing a gravitational function containing pose information.Finally,the feasibility of the scheme is verified by simulation experiments,and the direction of improvement is proposed through analysis experiments.