Multi AUVs Cooperative Navigation Based on Cross Entropy
Lichuan Zhang,Xiuye Tao,Feihu Zhang,Yijie Yuan
DOI: https://doi.org/10.1109/iccais.2018.8570451
2018-01-01
Abstract:Cooperative navigation can support the wide range navigation and positioning service for Autonomous underwater vehicles (AUVs), improving the positioning accuracy of the AUVs at the same time. In this paper, the Cross Entropy (CE) algorithm is applied to cooperative navigation system, in order to solve the problem of the path planning of the master AUV, and making the observation error of the system to be minimum, which would enhance the positioning accuracy of the AUVs. A slave AUV with low accuracy navigation system and a master AUV with high accuracy navigation system are used in the process of algorithm verification. First the navigation model is built in the framework of Markov decision process (MDP). Then, the CE algorithm is used to train the master AUV to select better path in the MDP navigation model. Finally, the optimal paths, which make the cumulative observation error to be minimum during the whole navigation process, are analyzed to verify the reliability of the CE navigation algorithm. The simulation results show that the CE navigation algorithm can select an optimal path of the master AUV to minimize the observation error.