Research On Real-time Visual Path Planning For Mobile Robots In Semi-structured Indoor Environment

Since the new century,the society and science are developing rapidly,people try to replace human labor by fully automated intelligent robots to serve human production and life.Handling robots and home intelligent service robots in various factories are at the forefront of research.To solve the problem of path planning in semi-structured unpredictable indoor environment in which a small part of unknown environment occurs while the basic environment of intelligent robots such as factory and family is fixed and unchanged is an unavoidable key issue in the research.The research on the path planning of mobile robots first appeared abroad in the1970 s,while the related research in China is later than that abroad.Up to now,the research in this field has yielded many results,but the results of the final vision have not yet been reached,so it is necessary to make unremitting efforts to do this work.Research.Early research on path planning of mobile robots was mainly based on the condition that the external environment information was fully known and known,that is,to find a feasible optimal or sub-optimal path from the starting point to the target point in a structured environment,which satisfies certain constraints.It did not set unpredictable obstacles to hinder the robot's progress,so it could only be set in a few specific artificial settings.When implemented in the environment,the adaptability to the environment is too low and the accuracy is limited.In recent years,research has begun to turn to a variety of algorithms and complex environments.New algorithms and improved algorithms are proposed to improve the accuracy.Meanwhile,the path planning of mobile robots hindered by obstacles is also discussed.However,even with the introduction of obstacles,many research institutes are still not deep enough,and the obstacles are simple,the environment is relatively specific,there are also characteristics of low adaptability to different environments.Based on the research of existing robot path planning algorithms,this paper proposes an improved adaptive genetic algorithm combined with the existing RRT*algorithm for path planning.By screening the descendants of each generation of population,a few of the most unsuitable individuals are eliminated directly,and themore real biological genetic characteristics are simulated.At the same time,the crossover rate and mutation rate are changed adaptively,which improves the yield based on traditional genetic algorithm.Because of the slow convergence speed and easy to fall into local optimum,the binocular camera is used to capture the surrounding information and transform it into two-dimensional grid image in real time during the mobile robot moving along the initial planning path.The RRT* algorithm is used to re-plan the mobile obstacle avoidance problem,by discuss and analyze the obstacles encountered on the way,then select the starting point and end point of local planning,and make corresponding obstacle avoidance decisions according to different situations,so as to solve the obstacle avoidance problem of moving obstacles encountered on the way,and make the planning path more effective and more adaptable to the changes of the environment.By analyzing the tasks and situations of multi-robot team in semi-structured environment,the information sharing process of multi-robot team is discussed,and the cooperative obstacle avoidance decision-making is made in the process of operation.This makes the path planning in semi-structured indoor environment more perfect.