Previous Abstract Return to Session C6 Next Abstract

Session C6: Collaborative and Networked Navigation

Research on Optimal Inter-vehicle Fusion Method for Low-cost-sensor-based Cooperative Navigation
Rong Wang, Zhi Xiong, Jianye Liu and Yuxuan Cao, Nanjing University of Aeronautics and Astronautics, China
Location: Windjammer
Alternate Number 2

In recent years, the research of Micro Air Vehicle (MAV) swarm is greatly focused for its potential and flexibility in various tasks, such as large area surveillance. In these tasks, navigation accuracy is critical for flight safety and task efficiency.
Accurate navigation of multiple MAVs in swarm is a challenge. An existing way is based on a ground station with a movement capture analysis system. But this way is heteronomous and limited by the distance from the station. Another existing way is utilizing the on-broad navigation system of each MAV independently. Nevertheless, due to the limitation of weight and power, the MAV usually equipped with low-cost sensors such as MEMS INS and GNSS receiver, whose positioning accuracy is unsatisfied for operation in swarm.
The cooperative navigation technology shows favorable prospect for improving the navigation performance of swarmed UAVs. With the development of networking communication and measurement technologies, cooperation among MAVs in swarm become practicable. Through communication network, the measurement data of different MAVs could be shared in swarm. Moreover, several technologies for inter-vehicle measurement, such as ultra-wideband ranging, were developed. The joining of the inter-vehicle measurement exploits information redundancy, which is beneficial for improving the overall positioning accuracy.
Compared with traditional mutual-independent navigation mode, the cooperative navigation mode brings changes in the characters of positioning uncertainty. In this paper, the distribution of the positioning confidence probability after introducing inter-vehicle measurement in swarm is modelled. From model analysis, it is concluded that the positioning uncertainty of a MAV would propagate to several MAVs’ positioning results in swarm. The mutual-coupled positioning uncertainty along the direction of inter-vehicle line follows a mixed Gaussian distribution. Meanwhile, the mutual-coupled positioning uncertainty along the normal direction of inter-vehicle line follows a symmetrical Gaussian distribution.
Based on the model analysis above, an optimal inter-vehicle fusion algorithm for cooperative navigation is proposed in the paper. The on-line modelling of the mutual-coupled positioning confidence probability is carried out according to geometrical configuration of MAVs in swarm. Then the expected value of the positioning offsets by the on-broad MEMS-INS/GNSS integration are obtained through optimal fusion by confidence probability. By theoretical analysis, it could be concluded that the compensation of the positioning offsets would reduce the uncertainty of positioning results, so as to improve overall navigation accuracy.
In the last of this paper, experiments have been undertaken to examine the validity and the potential of the proposed algorithm. The results shows that the above analysis and results are correct. The effectiveness of the proposed optimal inter-vehicle fusion algorithm in low-cost-sensor-based cooperative navigation is validated. In the situation of using customer-grade MEMS INS and GPS receiver, the proposed optimal fusion algorithm shows about twenty percent improvement in positioning accuracy over the traditional method. The experiment schemes, results and algorithms developed are provided in this paper.



Previous Abstract Return to Session C6 Next Abstract