Research On Localizability-based Path Planning Method For Mobile Robots

With the proposition of "Made in China 2025" action plan,the development of "Intelligent Manufacturing" has received the country's high attention and key support.Mobile robots(hereinafter referred to as "robots")are important research topics in this context and have a wide range of application prospects such as life services,smart logistics,disaster relief,and the like.Autonomous localization has always been one of the most basic and core technologies in robot-related technologies.Ensuring that robots have reliable localization performance in the autonomous navigation phase is a prerequisite for the robots to complete the assigned tasks.During autonomous navigation,environmental information and map noises at different locations may have dissimilar influence on the localization process of a robot.This implies that the robot's localizability,i.e.the ability to localize itself in an environment using laser range finder(LRF)readings,varies over a given map.Therefore,it is necessary to take localizability into account in the path planning stage in order to avoid large localization errors during navigation,and to reduce the chance of placing the robot at risk of failure to perform localization.In this thesis,the combination of localizability and path planning algorithm in the robot localization technology and its realization method are taken as the research objectives.A two-dimensional and three-dimensional localizability-based global path planning method for robots equipped with LRF sensors in the indoor and outdoor environment is studied and proposed,thereby enhancing the localization performance of robots navigating in texture-poor regions or regions with similar features or noisy maps,under GPS-denied environment.The main content of this thesis is as follows:1.Localizability matrix and localizability measure calculation:Based on Fisher's Infor-mation matrix,Cramer-Rao Bound theorem,previous work on two-dimensional LRF and probabilistic grid map(PGM),the need of path planning and the characteristics of mobile robots,estimation methods of localizability matrix and localizability measure estimation are proposed.More specifically,a two-dimensional localizability matrix and measure are calculated based on PGM and different kinds of two-dimensional LRF scan models;a three-dimensional localizability matrix and measure are calculated based on an OctoMap and different kinds of three-dimensional LRF scan models.The localiz-ability measure comprehensively reflects the localization uncertainty of the robots in each pose of a given map.2.Localizability-based path planning method:This thesis proposes a path planning method based on localizability(measure)constraint by introducing the localizabil-ity measure as a constraint during path planning phase.Based on the two-dimensional and three-dimensional localizability measure maps,the methods for extracting the local and global passable areas in the PGM and the OctoMap are given.The low localizability measure regions are filtered out in the constraint conditions to ensure that the robots possess satisfactory localization performance along the planned paths.3.Experimental verification:This thesis presents a series of real-world experiments and Gazebo-based simulations based on an intelligent wheelchair and a quadrotor model,compares and analyzes the experimental and simulation results accordingly.The results of multiple experiments and simulations altogether verify the feasibility and effective-ness of the proposed algorithm.In summary,aiming at the autonomous navigation task of robots,this thesis proposes a global path planning method based on two-dimensional and three-dimensional localizability for mobile robots.One of the advantages of this method is that it is not limited to any specific path planning algorithm in the optimization phase and can be easily realized by combining the localizability measure constraint with the traditional path planning algorithsm.A series of experiments conducted show that this method effectively reduces the localization error of the robot during the navigation process,and therefore improves the overall localization performance of the robots.