Collision-free Robot Motion Planning Based On Distance-labeled Grids

Many small-batch-and-diversity-product stations in manufacturing shops ever-increasingly demand increasing productivity and production quality,improving the automation and intelligence level of the manufacturing process,and reduce labor intensity.Along with these demands,human-robot collaboration has been one of the development tendencies.7-DOF collaborative robots,which have one redundant degree of freedom,have outstanding dexterity,and hence have the ability of working in small space or in an environment with obstacles.Therefore,the 7-DOF collaborative robots have attracted more and more attention and had more and more applications.In order to meet the demand for collision free when working in a small space or in an environment with obstacles,this thesis studies the inverse kinematics and distance-labeled-grid based collision free motion planning method for a 7-DOF robot.The results of simulations verify the proposed collision free motion planning method.The main work of this thesis contains: by virtual of analyzing the configuration features and the equivalent mechanism of a 7-DOF robot,this thesis introduces the redundant angle as an additional constraint when solving the inverse kinematics so as to the inverse kinematics solution can be determined uniquely;then by employing geometry method,this thesis obtains the closed-form solutions of the inverse kinematics of the robot;the planar obstacle discretization method is studied;the distance information of a grid,which is obtained by being scanned row by row,column by column,is considered as the grid's property,and then the 2D grid discretization and distance labeling are completed,and the distance information is saved as an matrix;the 2D grid discretization method is expanded to 3D space to deal with the 3D grid discretization and distance labeling;the distance information of the 3D grids is saved as a 3D matrix,which benefits the distance query process;the collision free motion planning is studied both at the robot end collision free path planning level and the robot body structure collision free planning level;the collision free path of the robot end is obtained by dichotomized vector searching method,so the high searching efficiency is achieved and the possibility of dropping into a local trap is eliminated;subsequently,given the robot end pose,to satisfy the demand for collision free,the elbow position,which is far away from the obstacle,is optimized by evaluating the distance of the grid in which the elbow center may appear;after that,the redundant angle corresponding to the elbow center position can be computed,and then the joint angles can be solved by employing the inverse kinematics method;thereby both the robot end and the robot body structure collision free motion planning are accomplished.At last,a typical application scenario is selected to verify the proposed methods.According to the given environment and task,discretize the target region,assign the distances,and plan the collision free path,successively,and then the motion path and the corresponding joint angles are obtained;consequently,the proposed methods are verified by the results of the simulation.The main contributions of this thesis are as follows:(1)the distance between a 2D or 3D grid and an obstacle are stored as the gird's property in the form of a 2D or 3D matrix;the distance labeling is done in the grid discretization process through direct assignment,and no complex detection and computation is needed,so it has the advantages of simple,easy to use,and effective.The distance labels facilitates the obstacle avoid planning a lot.(2)base on the proposed distance-labeled grids,this thesis proposes an collision free path planning method by simple geometric operations and searching;the method benefit from the distance-labels,and hence is simple and effective with no complex computation;it provides the collision free motion planning a feasible method.