Research Of A Local Planner Based On Stereo And 2D Lidar

Today,the rapid development of intelligent mobile robots and autonomous driving has also driven the research boom of related algorithms.The local path planning algorithm is an essential part of the intelligent mobile terminal to reach the destination safely and quickly.The local path planning algorithm gives the mobile robot or smart car the ability to successfully reach a predetermined target and complete a given task in a dynamic environment.Local path planning algorithms usually need to be based on stable and accurate positioning in a dynamic environment,known global maps,and quickly sense the dynamic changes of the environment as inputs,as well as the processing and decision making of integrated information.Therefore,the main research contents of this thesis include: localization algorithm in dynamic environment,and improved DWA(Dynamic Window Approach)algorithm combined with image analysis,and proposed visual environment-based dynamic environment localization and loosely coupled local path planning system based on 2D laser radar.The specific research work is as follows:First,it expands the positioning ability of traditional visual positioning algorithms based on static environment assumptions in dynamic environments.Combining the ability of deep learning image segmentation,screening potential moving objects in the environment,using traditional image processing algorithms to repair the results of deep learning image segmentation,and finally predicting the motion of potential moving objects,and participating in positioning in the image.The moving object is partially removed.The algorithm of dynamic object recognition and culling is applied to the visual localization algorithm.The experimental results show that the stability and accuracy of the original visual positioning algorithm are improved in the dynamic environment,and the performance of the world's most advanced dynamic environment localization algorithm is also similar.Upgrade.Second,a DWA algorithm combining image analysis information is proposed.Obstacle analysis is performed on the acquired image information,motion information of the dynamic obstacle is calculated in the image,and the dynamic grid in the grid map is marked,and the obstacle path prediction is added to the DWA path selection part for constraint.The DWA algorithm based on image analysis is applied to dynamic environment path planning and path computation.The experimental results show that in the dynamic environment,the DWA algorithm based on image analysis can select the safe path for the dynamic change of obstacles.Finally,the mobile robot and dynamic environment model are built on the simulation platform.The experimental research and effect evaluation of the research contents proposed in this paper are carried out,and the improved algorithm and the original algorithm effect are compared.The experimental results show that the new local path planning algorithm is more adaptable to the dynamic environment and takes less time than the original algorithm.