Carrier-Phase-Based Multi-Vehicle Cooperative Positioning Using V2V Sensors

This work proposes a new cooperative architecture that using Global Navigation Satellite System (GNSS), Inertial Measurement Unit (IMU) and vehicle-to-vehicle (V2V) observations to obtain robust and accurate inter-vehicle state estimation. A new cascade structure of relative filter which consists of float estimator and fixed estimator is presented that can take advantage of both the multi-sensor data and the information from Least-squares ambiguity decorrelation adjustment (LAMBDA). Also, a cooperative baseline estimation method based on multidimensional scaling (MDS) is proposed to further exploit the relative estimation from many other collaborators. Lastly, we combine the cascade relative filter (CRF) with MDS to estimate the relative states cooperatively using a feedback scheme. In the verification part, we use realistic sensor noise and a GNSS signal simulator to obtain inter-vehicle and GNSS measurements for a multiple-vehicle network. In a harsh GNSS scenario, only 51.20% of epochs in RTKLIB software can pass the LAMBDA acceptance test, but our proposed methods can achieve 77.85% (CRF) and 85.05% (CRF/MDS). Referring to the recovery time from a float solution to fixed solution, RTKLIB needs 179.23 s in the case of 4 satellites, but only 34.88 s (CRF) and 21.39 s (CRF/MDS) for the proposed methods. Results show that the proposed CRF has good performance when fusing with IMU and V2V observations, and has a better performance than existing methods. Moreover, the proposed architecture that combines CRF and MDS can have a further improvement, which substantially increases the robustness and accuracy of relative state estimation.