Research on Driverless Vehicle Positioning Based on Simultaneous Localization And Mapping in Low Visibility Environment

Abstract

Positioning is the key foundation of driverless technology. The existing Visual-Inertial Simultaneous Localization and Mapping (VI-SLAM) is difficult to extract a sufficient number of feature points in an outdoor low-visibility environment, resulting in a decrease in positioning accuracy or even positioning failure. GNSS satellite positioning is obviously complementary to VI-SLAM in this type of environment. Taking the fusion positioning method based on VI-SLAM and GNSS as the research object, this paper analyses its working mechanism, and proposes an LVG_SLAM algorithm including RFAST low-light image enhancement and GNSS fusion positioning module. The experimental results on public data sets and night road scenes show that the algorithm can effectively improve the positioning accuracy and robustness of the system under low visibility conditions.