Design And Implementation Of SLAM For Home Service Robots Based On ROS

In recent years, with the development and the maturity of mobile robots' hardware market, it is possible for the wide application of the mobile robots, especially with the robots popularization in various fields, the working environment for mobile robots is becoming more and more complicated and the requirements that the mobile robot can independently complete the task become higher and higher. As a result, the mobile robot autonomous navigation technology, which is the key for mobile robots to autonomous navigation, has gradually become a hot research topic in research area. In static environment, robot navigation technology has become more mature and obtained many research achievements under numerous specialists and scholars' joint effort. However, in unstructured dynamic environment, which is much like the real environment, there is no mature theory system and reliable solutions until now.Domestic researches of various research institutes on autonomous navigation technology of mobile robots are still independent from each other, so the original developed mobile robots' autonomous navigation system is facing the 2nd development under the circumstance that the robots' software and hardware system changes. In order to improve code reuse rate in robot research and development, this thesis focus on two key research problems in navigation technology under unknown dynamic environment with open-source ROS as software platform and simple ROS robots which autonomously designed and constructed by Jurassic work as hardware platform: Simultaneous Localization and Map Building.First and foremost, this paper studies several system models which is relevant with the SLAM, including: coordinate system model, the model of robot pose, odometer model, sensor measurement model and environment model. They lay the necessary foundation for the realization of algorithms on SLAM.Secondly, this thesis deduces the mathematical model for SLAM algorithms.Furthermore, we study thoroughly in these two kinds of SLAM algorithms, the extended Kalman filter and the particle filter.Then, analyze the performance of the above two kinds of SLAM algorithm in practical application, due to the respectively limitation the extended kalman filter and particle filter,this paper applied Fast SLAM which is a fusion of extended kalman filtering and particle filtering. And through the simulation, the environment map from Fast SLAM will be verified in simple navigation system which is built in Jurassic laboratory.This thesis adopts particle filter SLAM algorithm to establish laboratory environment real scene map so we can complete robot simultaneous localization and building task andThis thesis adopts particle filter SLAM algorithm to establish laboratory environment real scene map so we can complete robot simultaneous localization and building task and verify in the autonomous navigation. By conducting experiments, we validate the feasibility and efficiency of SLAM.