Empirical Study Of Laser Navigation And SLAM Algorithm Based On ARM Platform

In recent years, with the rapid development of autonomous robot technology, SLAM (Simultaneous Localization and Mapping) has become a hot topic in the field of intelligent robot research. The ARM processor itself located in the embedded platform, handling with lightweight and single-purpose program,is handy to be applied to mobile computing. The lidar is originally used in the military field, and has been extended to the field of unmanned vehicle. Due to the low power consumption of ARM processor and high accuracy of laser radar image,the research of SLAM algorithm for ARM platform is of great significance.In this paper, the key technologies involved in laser navigation are summarized, and four current mainstream laser SLAM algorithms are compared and summarized, in order to select the appropriate SLAM algorithm applied to embedded platform. Considering real-time embedded platform and the complexity of operating environment,compared with other methods, laser SLAM algorithm based on information fusion model has characteristics of simple calculation, high efficiency, no special requirements on data and easily dealing with multi sensor data, therefore choosing information fusion model as the foundation and direction of improvement. In this paper, the information fusion model is used as the main method to conduct research, regarding lidar data processing and particle filter principle as the breakthrough point to improve laser SLAM algorithm. The operational parameters of data processing includes choices of optimal environmental characteristics and particle population, along with lidar data pre-processing. The principle improvements of model includes integrating scan-matching technique and graph optimization theory into the information fusion model, in order to remedy predicted defect of unobtainable environmental characteristics,expected to improve the positioning and mapping accuracy of information fusion model.Experiment indicates that the target-aimed optimal selection of appropriate environmental characteristics will effectively improve the positioning accuracy by 4 percent. Choice of optimal particle population has a significant effect on enhancing the mapping precision of the information fusion model. Combined with scan-matching technique and graph optimization theory, information fusion model shall effectively conduct real-time navigation and the accuracy of localization slightly improves.