A robust LMEDS-TDCP method for low-cost single-frequency BDS/GPS precision relative positioning

Abstract High-accuracy and robust positioning are crucial for autonomous vehicles in urban road environments. However, Global Navigation Satellite System positioning results often deteriorate or even become unavailable due to factors such as signal barriers, multipath effects, and peripheral slippage. To overcome this challenge, a robust least median of squares—time differenced carrier phase (LMEDS-TDCP) model is constructed to solve the frequent outliers. Firstly, the LMEDS is combined with the conventional TDCP to reduce the impact of empirical thresholds on the velocity estimation. Secondly, LMEDS-TDCP estimated velocity is used as a constraint to update the prior position information in the Extended Kalman Filter, which can improve both the accuracy of the float ambiguities and the ambiguity fixing rate. Both pedestrian and vehicle tests were conducted to evaluate the effectiveness of the proposed method. The results show that the LMEDS-TDCP model obtains the highest robustness in environments with multiple gross errors compared with other methods. In terms of positioning accuracy, the proposed method achieves a 28.5% higher ambiguity-fixing rate than the traditional single-frequency single-epoch ambiguity resolution technique, with the root mean square error for 3D position difference improved from 1.64 m to 0.30 m.