Research On Obstacle Avoidance Path Planning For Multiple Degrees Of Freedom Manipulator

With the introduction and implementation of “Industry 4.0”,industrial production has gradually become intelligent,“Smart Plant”,“Smart Production” and “Smart Logistics” are the three themes of “Industry 4.0”.As the “New Type” in the intelligent manufacturing industry,robots liberate human from repeated and heavy work little by little.Because manipulator is one kind of the most used robots in intelligent manufacture,the performance of manipulator is directly related to production efficiency and production safety.And obstacle avoidance is a key and difficult point in the manipulator field,so it is of great significance to study on obstacle avoidance path planning for manipulator.The following aspects of obstacle avoidance path planning are studied in this thesis:First of all,as the theoretical basis of this paper,the manipulator kinematics were analyzed and studied deeply.After mastering the structure of manipulator,the forward kinematics model of NAO robot manipulator was established by coordinate transformation and DH method.Secondly,a new mixed bounding box is proposed.After studying on common different bounding boxes such as sphere bounding box,axis align bounding box,oriented bounding box and K-Dops bounding box,the advantages and disadvantages of various methods were summarized.Based on this,a new collision detection method called collision detection based on sphere-ellipsoid mixed bounding box was raised.This method gives consideration to simplicity and tightness of bounding box at the same time,and it has fine modeling effects both on the convex obstacles and non-convex obstacles.Afterwards,the superiority of this method is verified by simulation experiments.Then,a modified artificial potential field(APF)is given for obstacle avoidance path planning.Firstly,intensive studies of common obstacle avoidance path planning of robot were made,considering merits and drawbacks of various kinds of methods,APF was chosen to proceed on obstacle avoidance path planning,this method has a simple structure,smooth generation path,lower hardware requirements,and application scope more extensive.Then,to solve the problem that using traditional APF might fall into local optimum easily,some improvement measures are put forward.Firstly,by appropriately modifying the potential field function,some situations that fall into local optimum are removed.When path planning falls into local optimum,virtual goal point will be added automatically which produced by RRT algorithm.The virtual goal point is used to change the environment around the manipulator.Using this method manipulators can jump out of local optimum quickly.Feasibility of this method is verified by simulation experiments.Finally,the method adopted in this paper is verified by physical experiments based on NAO robot platform,the results show that it can plan a collision-free path quickly and effectively for manipulator by the proposed methods.