Mobile Robots Collaborative Navigation Research Based On Multi-source Sensor Information Fusion

Nowadays,the integrated navigation system applied to a single moving body has certain limitations.The cooperative navigation system applied to multiple moving bodies can make up for the deficiency of individual ability which is the current research hotspot.This topic focuses on the information fusion process of multi-source sensors in mobile robot cooperative navigation aiming at improving the navigation accuracy and robustness of multi-mobile robot cooperative navigation system.The main research contents are as follows:Firstly,according to the situation that the mobile robot is equipped with different auxiliary navigation systems,different collaborative navigation models are established in combination with the actual working environment.They are odometer cooperative navigation model,odometer/GNSS coordinated navigation model,odometer/visual collaborative navigation model and collaborative navigation model in complex environment.Corresponding mathematical descriptions of the sensors involved provide a basis for subsequent research.Secondly,for the odometer/visual collaborative navigation model,the related SLAM data association algorithm is studied and applied to local map data association and loopback detection.Aiming at the cooperative navigation model in complex environment,the detection and suppression schemes of anomaly observation in the process of cooperative navigation of mobile robots are designed,which effectively improves the robustness of the collaborative navigation system.Then,according to the multi-sensor data fusion structure,the sensor data fusion framework of centralized and decentralized collaborative navigation is designed.Based on the proposed collaborative navigation model,the collaborative navigation solution method is studied.The direct relative observation and the indirect relative observation are studied,and the relative observation models are designed respectively.The correlation of the state information of each node in the collaborative navigation process is analyzed,and the concept of the separation covariance cross filter is introduced.Information coupling,and finally proposed a decentralized collaborative navigation algorithm.Finally,based on the established collaborative navigation model,different simulation experiment scenarios are designed,the centralized and decentralized cooperative navigation algorithms are verified,and the navigation results of mobile robots are compared and analyzed.The centralized and scattered experiments in different experimental scenarios are discussed.The pros and cons of the collaborative navigation algorithm.