Near-Field Localization of Mobile Robots with Multiple Access Points Collaboration

Utilizing unequal confidence angle of arrival (AoA) measurements from multiple access points (APs) in a network can significantly enhance the localization accuracy of robots. However, with the large-scale increase of antenna arrays, the near-field region with both AoA and range information becomes non-negligible. Existing confidence quantization schemes, which are based on far-field channel assumptions, fail to account for the estimation accuracy of both AoA and range. To this end, in this paper, we propose a near-field localization architecture for mobile robots that incorporates multiple APs with unequal confidence levels. Specifically, we first develop a near-field localization model to directly estimate AoA and range parameters by integrating multi-antenna channel state information with the robot's trajectory. Second, we introduce a confidence quantization scheme for multi-AP measurements that combines spatial spectrum measurements with multipath interference assessments and eliminates outliers. Additionally, we design weighted least squares localization schemes tailored to the spatial-spectral characteristics of different trajectories. Finally, we implement a prototype experimental system using a mobile robot and commercial WiFi devices. Experimental results demonstrate that the proposed system operates effectively in complex environments and achieves superior localization accuracy compared to existing methods