Research On Fusion Positioning Algorithm Based On Ultra-wideband And Visual SLAM

The rapid progress of high-tech not only improves social production efficiency but also changes human life,and the increasingly commercialized robots in recent years are a major manifestation of using high-tech to improve daily life,and simultaneous localization and mapping(SLAM)is the core technology.The research has extremely important exploratory significance for SLAM positioning technology with strong practicality and higher positioning accuracy in application scenarios.Localization is an inseparable part of SLAM technology,how to better adapt to the unknown environment of unstructured or featureless areas and reduce the trajectory drift accumulated by long-term motion is still a key issue in visual SLAM positioning technology.Monocular visual positioning is the most widely used and the hardware cost is the lowest in SLAM,but there is a scale error in monocular visual odometers,that is,scale uncertainty.For the above problems,many scholars use multi-sensor fusion to solve.In this paper,a fusion positioning technology scheme using Ultra-Wide Band(UWB)and visual monocular SLAM positioning technology is proposed.In this paper,the tight fusion method combining the advantages of UWB sensor and monocular camera is used for positioning,the distance measurement information obtained by a single ultra-wideband anchor point UWB is used to correct the camera posture information estimated by the monocular vision method,and the UWB ranging data obtained between a static anchor placed in the environment and the UWB tag on the robot is used to propose the fused state variable and optimize it using the Levinberg-Maquart algorithm to derive the scale factor and the anchor position to be estimated.There is no need for prior estimation of anchor points,taking advantage of the complementarity of different types of data.Because the UWB sensor is able to cope with challenging vision technology problems,overcoming the problem that pure vision methods rely too much on the geometric information of the environment,the centimeter-level positioning of the robot in the unstructured environment is completed.In order to further improve the efficiency of estimation optimization and reduce unnecessary computing resource consumption,a double-layer sliding window is used to provide accurate posture tracking for data optimization,and UWB ranging data is used as a constraint,which greatly reduces the complexity of computing,and this method is particularly effective on low-power computing platforms.In order to verify the effectiveness and practicability of the ultra-wideband UWB and monocular camera SLAM fusion positioning algorithm proposed in this paper,the positioning simulation experiment is carried out on the publicly available authoritative Eu Ro C dataset,and the positioning results are evaluated and analyzed.Experimental results show that the proposed algorithm is accurate and effective,and can provide more accurate location information,with good practicality.