Study On Mapping And Path Planning Of Overhead Traveling Crane

As the most widely used lifting machinery,overhead traveling cranes still have poor levels of operation efficiency and automation.This thesis focuses on the automatic operation of overhead traveling crane,studies the construction of indoor off-line three-dimensional point cloud map based on RGB-D SLAM,two-dimensional and three-dimensional map representation for workshop environment,and the global path planning of overhead traveling crane based on improved ant colony algorithm in two-dimensional plane.Firstly,the models such as coordinate systems model,kinematic model,sensor observation model are built for the system of overhead traveling crane visual SLAM,and then completed the calibration of Kinect.Visual odometer is built by image feature extraction and matching based on ORB algorithm and BF algorithm and camera motion estimation based on ICP algorithm,and on this basis a dense map of the environment is constructed by back-end based on graph optimization using g2 o tool and loop closure detection based on random detection method.2D and 3D grid-based map are obtained by the processing of points cloud filtering in PCL and octomap libraries.Secondly,an improved ant colony algorithm for the path planning is proposed on the basis of the study of ant colony algorithm.Aiming at the shortcoming of ant colony algorithm in slow convergence speed and falling into the local optimum easily,improved the adaptive heuristic function and the pheromone-updating mechanism borrowed A* algorithm and the assignment rule of wolf colony.According to the operating characteristics of the overhead traveling crane,path length and number of nodes are abstracted as simulation factors by kinematics and dynamics analysis.A new evaluation criterion at the cost of performance parameters such as path length,running time and stability is proposed.The simulation results for path planning in grid environment show that improved algorithm improves the convergence speed,avoids local optimum and can obtain better path applied to engineering.Finally,point cloud map of experimental platform is obtained by RGB-D SLAM,and 2D grid-based map is obtained by the processing of points cloud filtering,then,the global path is planned by improved ant colony algorithm based on 2D grid-based map.The running results of the experiment platform show that the hook can avoid obstacles to reach the destination along the planned path,which verifies the feasibility of the scheme.