Decentralized cooperative navigation solution for a swarm of UAVs operating in GNSS degraded environment

This paper presents a decentralized cooperative navigation algorithm which can prevent the overall state estimation divergence by providing a good estimate of cross-covariance terms also in architectures which envisage fully connected graphs. i.e., all the intra-agents measurements are used to update the state of the formation. The proposed approach assumes each agent can independently use non cooperative measurements to update its state. Conversely when cooperative measurements are available a platform (which is iteratively chosen among the formation) is selected to collect all the cooperative information, perform the update and communicate the updated state and covariance to its fellows. The cooperative process is innovative with respect to the other works in the open literature which either make empiric approximations so that cross-covariances can be overestimated or perform cooperative updates by iterating on pairs chosen among the available platforms, thus not allowing an exact covariance update. The proposed algorithm is tested on a scenario simulating fixed wing aircrafts and assuming ranging measurements as intra-agents’ aid. Other simulated measurements include GNSS, IMU, magnetometer and ranging information with respect to moving landmarks. Results are shown by comparing the performance of the proposed algorithm with the centralized (and thus optimal) architecture. In addition, algorithm working conditions are stressed by considering either platform failure or GNSS measurements and other positioning sources (i.e., moving landmarks) occlusions. The latter condition has been considered to test the behavior of the navigation algorithm when only intra-agents ranging are available to bound the formation’s overall navigation error. Results demonstrate the algorithm to perform similar to centralized solution, by mostly depending on the geometry on which the moving landmarks are placed.