Research On Slam Method For Bio-inspired Polarized Skylight Integrated Navigation System Based On Graph Optimization

Simultaneous Localization and Mapping(SLAM)refers to the process of constructing environment map synchronously when a moving object calculates its own position and pose according to the sensor measurements from the environment.As a key technology for autonomous mobility and sensing of robots,SLAM has attracted wide attention.It has been widely applied in aerospace,vehicle transportation,logistics,warehousing and other fields.At present,how to realize the localization and mapping with high precision,full autonomy and excellet anti-jamming capability has become a hot topic,Aiming at the sloving the problems of low autonomy,pool anti-jamming ability and low precision in traditional SLAM methods,a bionic polarized skylight sensor with strong autonomy and no accumulated error is introduced to provide a heading angle.The bionic polarized skylight sensor is used to integrated the Inertial Navigation System(INS),lidar and optical flow to constrcut a complemenatry navigation systems.A SLAM for bionic integrated polarized light navigation system based on graph optimization is proposed.A bionic polarized skylight multi-sensor fusion SLAM method based on filter and graph optimization was studied,respectively.Finally,the SLAM with high accuracy,full autonomy and good anti-interference was obtaiend.The main contents of this paper are given as follows:1.In order to solve the problems of the traditional INS/Global Navigation Satellite System(GNSS)integrated navigation system,such as lack of autonomy and easy to be interfered by electromagnetic signals and blocked by buildings,the bionic polarized light sensor and optical flow sensor are introduced to assist INS.The heading measurement equations of the bionic polarized skylight sensor and the velocity measurement equations of the optical flow sensor are established,respectively.An integrated navigation system based on INS/polarization/optical flow is established,and an autonomous navigation method for hexapod robot based on INS/polarization/optical flow is proposed.Simulation and experiments show that the integrated navigation system does not need to receive external navigation signal,and it can effectively reduce the cumulative error of attitude estimation,and has high real-time performance and full autonomy.2.Aiming at slove the problems of low precision and insufficient autonomy of traditional multi-sensor integrated SLAM system,the bionic polarized skylight sensor was introduced.The heading angle measurement equation based on the bionic polarized skylight sensor was established,and a multi-sensor fusion SLAM method based on INS/polarization skylight/odometer/lidar was proposed.In order to analyze the effect of bionic polarized light sensor on the improvement of pose estimation accuracy of SLAM system,the observability analysis of INS/polarization skylight/odometer/lidar combined system was carried out.At the same time,the federated error-state Kalman filter(ESKF)was designed to realize the efficient integration of multiple sensors including INS,bionic polarized light sensor,odometer and lidar.Experimental results show that the proposed method can effectively improve the position accuracy and greatly reduce the heading error.It can be applied to the actual outdoor scene.3、Aiming at the solving the problems of low accuracy of SLAM in sparse scene,which is greatly influenced by outliers,a polarized camera is introduced.The measurement equation of polarizing camera is established to eliminate the influence of environment outliers,which is combined with lidar and INS,a factor graph based on INS/Lidar/polarized camera is constructed,and the optimal state estimation is obtained by using the method of factorial graph optimization.Experimental results show that this method can effectively improve pose accuracy and robustness to outliers.