Research On SLAM Algorithm Of Lunar Roving Vehicle

Autonomous navigation technology is the fundamental guarantee for Lunar Roving Vehicle(LRV)working steadily in an unknown environment,localization and mapping are two key problems for this technology,which are correlated.Combining visual and inertial measurements is the ideal choice for accurate localization and mapping of LRV,since the two sensing modalities offer complementary characteristics.This paper performs intense research about Visual-Inertial Simultaneous Localization and Mapping(SLAM)problem for LRV,mainly includes the following aspects:Firstly,the Visual-Inertial system model in the study of the SLAM problem is established for the three dimensional environment,including the kinematic model of the Inertial Measurement Unit and the stereo camera observation model,which lays the theoretical foundation for the algorithm design.Secondly,as the lunar lighting conditions are poor and lunar surface is an unstructured environment,a feature detection and extraction algorithm with fast detected speed,less space and fast computation is studied.By doing experiment to verify the properties of the algorithm at the correct matching rate and computation time.Thirdly,the noise in SLAM problem is solved by non-linear optimization,which improves the accuracy of pose estimation and map points positons.At the same time,using marginalization to control the number of optimization variables,in order to guarantee real-time using of our algorithm.In addition,the map of the surrounding environment is built,which consists of sparse feature points and the trajectory of LRV.Finally,the simulation experiment of Visual-Inertial SLAM algorithm is carried out by using the open datasets.The results show that the proposed algorithm has high accuracy and can reconstruct the surrounding three environment.