Robot Indoor Positioning Method Based on Multi-Sensor Fusion and Improved Algorithm

近年来，随着室内无人定位技术的发展，市场对无人智能系统的需求日益增长。然而，室内环境的复杂性也给定位技术带来了不小的挑战，由于空间限制，物体之间常常会互相遮挡或干扰，这不仅严重影响了传感器的识别能力，还降低了定位的准确性。为了解决这一问题，本文提出了一种多传感器融合的方法。该方法以激光雷达(LiDAR)为主传感器，同时辅以深度相机、WiFi和惯性测量单元(IMU)来增强系统的性能。针对由障碍物遮挡引起的点云匹配不准确的问题，本文设计了一种改进的点线迭代最近点(Point-to-Line Iterative Closest Point, PL-ICP)算法，该算法在遮挡场景下能够显著提高匹配的精度和速度。此外，本文还对Otsu算法进行了改进，使其能够更好地利用深度相机采集到的图像(RGB-D图像)匹配来提取额外的特征信息，从而在存在遮挡的情况下增强系统的收敛性。最后，采用扩展卡尔曼滤波器(EKF)算法来融合点云、图像以及WiFi定位数据，进一步提升了定位的准确性和鲁棒性。经过大量实验验证，本文的方法不仅提高了定位的精度和稳定性，还展现出了很好的收敛特性。这为机器人定位和导航提供了一种既经济又高效的解决方案。In recent years, with the development of indoor unmanned positioning technology, the market demand for unmanned intelligent systems is increasing. However, the complexity of the indoor environment also brings great challenges to the positioning technology. Due to space constraints, objects often block or interfere with each other, which not only seriously affects the recognition ability of the sensor, but also reduces the accuracy of positioning. In order to solve this problem, this paper proposes a multi-sensor fusion method. This method uses LiDAR as the main sensor, supplemented by depth camera, WiFi and inertial measurement unit (IMU) to enhance the performance of the system. Aiming at the problem of inaccurate point cloud matching caused by obstacle occlusion, this paper designs an improved Point-to-Line Iterative Closest Point (PL-ICP) algorithm, which can significantly improve the accuracy and speed of matching in occlusion scenes. In addition, this paper also improves the Otsu algorithm so that it can better use the image (RGB-D image) matching collected by the depth camera to extract additional feature information, thereby enhancing the convergence of the system in the presence of occlusion. Finally, the extended Kalman filter (EKF) algorithm is used to fuse point cloud, image and WiFi positioning data, which further improves the accuracy and robustness of positioning. After a large number of experimental verification, the method in this paper not only improves the accuracy and stability of positioning, but also shows good convergence characteristics. This provides an economical and efficient solution for robot positioning and navigation.