面向地下空间探测的移动机器人定位与感知方法

提出了一种面向地下空间探测的移动机器人定位与感知方法。首先，针对地下空间的结构退化问题，构建了基于因子图的激光雷达／里程计／惯性测量单元紧耦合融合框架；推导了高精度惯性测量单元／里程计的预积分模型，利用因子图算法实现对移动机器人运动状态及传感器参数的同步估计。同时，提出了基于激光雷达／红外相机融合的目标识别方法，能够对弱光照环境下的多种目标进行识别与相对定位。试验结果表明，在结构退化环境中，本文方法能够将移动机器人的定位精度提升50%以上，并对弱光照环境中的目标实现厘米级的相对定位精度。

Mobile Robot Localization and Perception Method for Subterranean Space Exploration

A mobile robot localization and perception method for subterranean space exploration is proposed. Firstly, a tightly-coupled LiDAR-odometer-IMU (inertial measurement unit) fusion framework based on factor graph is designed for the problem of structural degeneration in subterranean space. A high-precision IMU/odometer pre-integration model is derived, and the factor graph is utilized to achieve the simultaneous estimation for motion states and sensor parameters of mobile robots. Meanwhile, an object detection method based on LiDAR and infrared camera is proposed to conduct the recognition and relative localization of multiple objects in weak light environment. The experimental results show that the proposed method can improve the positioning accuracy of mobile robots by 50% in structurally degenerated environments and achieve a centimeter-level relative localization accuracy for objects in weak light environment.