Study On Localization And Path Planning For An Indoor Mobile Service Robot Based On Laser

Mobile robots provide a lot of conveniences for human production and life.In recent years,many service robots appears in various fields.Localization and path planning become necessary foundations for intelligent service robots.At present,localization methods mainly include visual and laser.Since the cumbersome steps,complex calculations and high requirements for configuration,the cost of visual localization is high.Many enterprises and research institutions tend to adopt accurate and stable lasers.Commonly used laser localization methods generally fuse with other sensors or extract environment characteristics and the calculations are complicated.The common path planning methods have the problem of computation burden.In the case of obstacle avoidance,they consider all the surrounding areas of the robot,which are in short of pertinence.Aiming at this situation,this paper mainly provides solutions for low cost and low configuration mobile service robots.The dissertation analyzes the robot kinematics to control the robot.The data filtering and environmental modeling methods are proposed.The localization of the robot is achieved through point cloud registration and the features of ICP are tested.Above this,the registration method based on FFT is proposed.Converting the effective laser data to grid matrices,using theories of FFT and integral projections to solve registration.The method is verified by experiments.Based on this,the dissertation adopts FFT-ICP method to locate the robot.The experiments are carried out,which show that the method not only can be applied to the case of low point cloud overlap,but also can realize precise positioning of the robot at different moving speeds.Using binary grid method to establish the environmental model,the global path planning is carried out using the improved PS-A* algorithm to enhance practicability.In order to solve the obstacle problem of local path planning,three kinds of environmental perception models are proposed and analyzed.We choose the curved quadrilateral model to build local maps.The obstacle avoidance strategy based on local maps is proposed.We present a motion controller to achieve the robot trajectory tracking control.The simulation software of path planning is designed,the proposed path planning methods are simulated.The results verify the effectiveness of the methods.The actual tests are carried out.The dynamic localization experiment and global experiment prove that the proposed methods can be used to locate the robot in real time and the robot has a good path planning capability.