Distributed Dynamic Consensus Algorithm for Attitude Estimation

This paper proposes a new dynamic consensus algorithm for the attitude estimation of a rigid body using n Measurement Units (MUs) spatially distributed on the body. The MUs are considered as agents in a sensor network, which is modeled using graph theory. It is assumed that the nth MU shares attitude information only with its neighbor agents. Then, a dynamic consensus algorithm is developed to obtain an average quaternion that describes the general attitude of the rigid body. The objective is to get a more accurate and reliable estimation of the attitude, further providing a fault-tolerant system in the event of failure sensors. This new dynamic consensus algorithm is validated in simulation using three MUs in the network. Results show a better precision and demonstrate that the proposed algorithm provides a good estimation of attitude even in the scenario of possible failure of any of the sensors.