Research On Positioning And Nagation Technology Of Mobuile Robots Based On SLAM Algorithm

In recent years,with the rapid development of artificial intelligence technology,mobile robots have been widely used in many fields.In various unknown environments,autonomous positioning and navigation of mobile robots is becoming more and more important.Positioning and navigation technology is an important prerequisite for realizing the various functions of mobile robots.At present,there are many research achievements in the positioning navigation of mobile robots in the known environment.However,how to realize the positioning and navigation of mobile robots in an unknown environment is a hot topic in current mobile robot research.The SLAM(simultaneous localization and mapping)algorithm can realize the on-line measurement of the environment information and the estimation of its own position by the sensor carried by the robot.It provides an effective solution for the positioning and navigation for the mobile robot in the completely unknown environment.Therefore,research on positioning and navigation technology for mobile robots based on SLAM algorithm has important academic significance and research value.In this thesis,the principle of SLAM algorithm is analyzed in detail,SLAM algorithm problem can be transformed into solving mobile robots motion model and observation model.Further,the low accuracy of the traditional SLAM algorithm is considered.The extended Kalman filter SLAM algorithm based on time-varying adjustment factor and Fast SLAM algorithm based on artificial fish are presented respectively.Then,based on the presented SLAM algorithms,the environmental information acquisition and processing system of mobile robots is designed and implemented.The main research content in this thesis is as follows:Firstly,the SLAM algorithm can be considered as a problem of probability estimation by analyzing the basic principles of the SLAM algorithm.The optimization problem of the SLAM algorithm for mobile robots is to make the estimated robot pose and the position information of the road marking points as close as possible to the actual pose and signpost points.Therefore,the mobile robot motion model and observation model are developed.The obtained models provided an effective research platform for the subsequent optimization research of the SLAM algorithm.Secondly,an EKF-SLAM algorithm based on a time-varying adjustment factor is presented for the disadvantages of the traditional Extended Kalman Filtering SLAM(EKF-SLAM)algorithm,such as low accuracy and poor performance.The filter gain can be adjusted on-line by introducing a time-varying adjustment factor in order to improving the estimation accuracy.Experimental results show that the presented EKF-SLAM algorithm based on the time-varying adjustment factor makes the estimated state estimation of the system closer to the true value.Also,the presented method improved position estimation accuracy of the mobile robot.Thirdly,the traditional Fast SLAM algorithm estimate mobile robot pose by particle filter,which often leads to the deviation of the particle set sample from the robot real pose.Therefore,a Fast SLAM algorithm based on artificial fish swarm optimization is proposed.The artificial fish swarm algorithm is used to optimize the estimated particle set and adjust the proposed distribution of particles so that the predicted particle sample is closer to the true distribution before the weights are calculated.Also,the predicted particles can be as close as possible to the prediction of the real system state distribution.Therefore,the estimated state of the system is closer to the actual value.Simulation experimental results show that the proposed method can effectively improve the robustness and estimation accuracy of the traditional Fast SLAM algorithm.Finally,a kind of environment information acquisition and processing system is designed for the mobile robot SLAM algorithm based on DSP chip TMS320F2812.The information of the road signs which in the environment can be acquired by mobile robots vision sensors.Also,the position of the landmarks can be estimated by feature extraction and feature matching methods,the results can be used as the observation model in the SLAM algorithm.The pose of the mobile robot is estimated by an encoder and used as a motion model.The observation model and the movement model are used as the input of the filter,and an incremental construction environment map is carried out to realize the simultaneous positioning and map construction function of the mobile robot.Experiments have further verified the effectiveness of the environment information collection system based on the presented SLAM algorithm.