MSFusion: Multilayer Sensor Fusion-Based Robust Motion Estimation

Vision-based measurement has become a research hotspot for autonomous vehicles, but most works ignored the information from other perspectives, and its robustness and accuracy still need to be improved. In this letter, we proposed a novel multilayer fusion framework to achieve front-rear camera arrangement-based multisensor fusion localization using a nonlinear optimization method. In this work, we have designed a factor to evaluate the effect of optical flow tracking, realized the complementarity of front and rear visual-inertial odometry, and proposed a bidirectional loopback strategy, thereby improving the robustness and flexibility of the system. The performance of our method has been evaluated on open source dataset and in the real world, and it is demonstrated that our method outperforms other state-of-the-art algorithms.