Research Of Vision-guided Obstacle Avoidance Path Planning Method For Redundant Space Manipulator

The path planning problem is a significant branch in the field of robotic research,and also one of the most important steps in robotic navigation tasks.Due to the complexity of space environment,it is necessary to analyze the kinematic flexibility of manipulator in global path planning methods,to advance kinematic performance while avoiding obstacles and to improve the capacity of handling unexpected emergency.Therefore,research on the obstacle avoidance path planning methods of redundant space manipulator has both theoretical and modern significance,also broad application prospect.Redundant space manipulator is set as the research object in this paper.The construction method of environment model based on stereo vision technology,the analysis of kinematic performance in obstacle avoidance process,the RRT optimization algorithm guided by independent potential field and its simulation experiments are carried out in this paper.The main research work is as follows:Firstly,a construction method of environment model based on stereo vision technology is carried out.The disparity images of target scenes are obtained by stereo matching algorithm based on semi-global block matching technology.Then,the method to obtain point clouds model with dimension reduction is studied.In this method,an entropy method combined with principal component analysis is used to do the dimension reduction of AKAZE feature descriptor,to improve the efficiency of AKAZE feature matching process.The point clouds model about environment is obtained by feature matching,PnP algorithm and bundle adjustment.Finally,an octomap is built by the modeling method based on octree,which establishes the foundation for solving obstacle avoidance path planning problem.Secondly,the analysis of kinematic performance in obstacle avoidance process is studied.In order to improve the capacity of handling unexpected emergency,reduce the operational cost and extend the service life of manipulator,the relationship between its kinematic performance and the kinematic indexes,such as the minimum singular value of Jacobian matrix,manipulability,condition number,joint limits of manipulator is established.And the optimization objectives in obstacle avoidance process are determined,which are used to build the potential field.Finally,a RRT optimization algorithm guided by independent potential field and its simulation experiments are carried out.A RRT optimization algorithm guided by independent potential field is proposed.Based on the characteristics of global path planning algorithm,the potential field is established by those selected optimization objective in configuration space.The potential field method is used as optimization method,to attract manipulator into some configurations or repel it from some configurations,to optimize the kinematic performance of manipulator.The potential field is only depend on its own structure rather than specific tasks.So the potential field only needs to be built once if the structure of manipulator stay the same.The RRT is combined with the independent potential field built offline.The potential field is using to guide the RRT to extend,which make the manipulator have better kinematic performance while obstacle avoiding.A KNN based collision detection strategy is integrated into our algorithm to improve planning efficiency further.In the end,a visual simulation experiment platform based on VS2013 and Matlab2016b is established to do the experiment.The result shows the algorithm can effectively plan a barrier free path.Moreover,the manipulator can always maintain a safe distance from.obstacles and joint limits.The planning efficiency is also maintained.From the above,our algorithm can reduce the manipulator cost and to improve its kinematic performance.