毫米波雷达辅助的多机协同定位与建图研究

Research on millimeter-wave radar-assisted multi-agent collaborative localization and mapping

随着无人驾驶技术的发展,低空物联网逐步成为国家产业创新的重点方向。高精度与强适应的实时定位与建图是实现无人系统智能化运行的关键。然而,单一无人系统感知能力的局限性以及视觉传感器的易受干扰性,致使传统方法难以满足复杂动态环境下的实时定位与建图需求。为提升定位与建图的准确性和适应性,创新性地提出了一种毫米波雷达辅助的多机协同定位与建图方案。该方案深入挖掘毫米波雷达时、空、多普勒域等多维度特征,有效筛选可靠点云信息实现本地里程计以及局部地图构建。基于共享局部地图信息,设计多无人系统协同的全局优化机制,实现精准定位与建图。实验结果表明,该方案有效保障了系统的鲁棒性和精确性。

With the development of autonomous driving technology,the low-altitude Internet of things has gradually be‐come a key focus of national industrial innovation.High-precision and highly adaptable simultaneous localization and mapping are crucial for the intelligent operation of unmanned systems.However,the limited perception capabilities of a single agent and the susceptibility of visual sensors to interference make it challenging for traditional methods to meet the demands of real-time localization and mapping in complex and dynamic environments.To enhance the accuracy and adaptability of localization and mapping,an innovative millimeter-wave radar-assisted multi-agent collaborative localiza‐tion and mapping scheme was presented.This scheme deeply explored the multi-dimensional features of millimeter-wave radar in the time,space,and Doppler domains,effectively filtering and collecting reliable point cloud information to achieve local odometry and local map construction.With shared local map information,a collaborative global optimiza‐tion mechanism was designed to achieve precise localization and mapping.Experimental results demonstrate that this scheme effectively ensures the robustness and accuracy of the system.