Decentralized Cooperative Navigation for Vehicle-to-Vehicle (V2V) Applications using GPS Integrated with UWB Range

This paper proposes a cooperative vehicle-to-vehicle (V2V) relative positioning approach using a federated filtering scheme in a Vehicular ad hoc Network (VANET) based on the techniques of differential GPS (DGPS) baseline estimation and Ultra-Wide Band (UWB) ranging. The relative positioning problem is addressed by assuming each vehicle in the VANET is equipped with a GPS receiver, UWB ranging radio, and communication devices through which information can be exchanged among elements in the network. The solution to this problem is a federated filtering structure where each vehicle runs a decentralized local filter estimating the relative positions between it and the neighboring vehicles using all local observations and a global fusion filter that fuses each vehicle's local filter estimate with the other vehicles' local filter estimates updated using only independent observations to this vehicle. The proposed approach relies on V2V communication that broadcasts each vehicle's local GPS observations for differential processing and each vehicle's local estimate to the other neighboring vehicles for further global fusion. A GPS and UWB data collection campaign involving three vehicles traveling around a "T" shape intersection was conducted to demonstrate the effectiveness of proposed approach. GPS pseudorange and Doppler and UWB range observations were used for V2V baseline estimation with 20-30 cm of accuracy achieved. The benefit of the proposed approach is assessed by comparing it with traditional moving-base station DGPS involving only two vehicles.