A Lightweight Real-time Multi-sensor Fusion State Estimation Method Based on Event Camera in Challenging Scenarios

State estimation refers to the integration of sensor input and observation information to estimate the current state of a robot, such as its position and orientation. However, achieving robust and accurate state estimation in challenging scenarios remains an urgent problem to solve. In this paper, we propose a novel Event-Visual-Inertial Odometry (EVIO) method that tightly couples event data, standard images, and inertial measurements. This method fully leverages the excellent complementary characteristics of multi-sensor fusion to estimate the ego-motion of sensors in challenging scenarios such as High Dynamic Range (HDR), low light, and motion blur caused by vigorous motion. In challenging benchmark experiments, our method outperforms state-of-the-art methods by 16%, and this achievement does not rely on extensive manual parameter tuning. Lastly, we qualitatively demonstrate the accuracy and robustness of our method for real-time state estimation across different scales and types of challenging scenarios.