A Structured Scene-Oriented SLAM System Based on Low-Cost Multi-Sensor Fusion

Abstract

Structured scenes are characterized by complex road conditions and poor GPS signals, and map degradation and degradation of positioning accuracy often occur when robots build maps of structured scenes. Aiming at the above problems, a low-cost multi-sensor-periodize fusion SLAM system is designed, which fuses four low-cost sensors, namely, 2D LIDAR, RBGD camera, inertial measurement unit, and wheel odometer. A sub-echelon data processing session is designed in the motion initialization session, and a sensor multi-strategy selection method is designed in the optimization and alignment session to maintain the robustness of map building and localization with less computational effort for optimization. Compared with the traditional method, it can improve the positioning accuracy and constrain the map degradation rate to less than 5%, which substantially improves the value of practical engineering applications while meeting the requirements.