A Distributed Cooperative UAV Swarm Localization System: Development and Analysis

摘要

*ION GNSS+ 2017 Student Paper Award Winner* Cooperative swarms of robots have been proven to be fast, efficient, and robust systems for performing tasks such as mapping, exploration, search and rescue, surveillance, disaster management etc. Hence, several researchers are investigating the use of swarm networks for aerial, land, and marine applications. A key requirement in all applications involving swarms is localization of each of the nodes. This paper develops a distributed cooperative swarm localization system using GNSS, inertial and Ultra-Wide Band (UWB) sensors. The system is distributed since each of the nodes perform computations locally i.e. no central server is involved. The localization system is cooperative since each of the nodes in the swarm shares their information with all the other nodes and thus assists nodes without GNSS in achieving robust localization. The major contributions of this paper are development and analysis of a cooperative localization system. This paper presents the first results of this distributed system, provides a performance analysis of the localization system, and highlights the challenges in developing a true robust distributed cooperative localization system. To the best of the author’s knowledge, this is the first attempt in realizing a prototype of a true distributed cooperative localization system using low cost sensors. The experimental results show that in the absence of GNSS, it is possible to achieve an accuracy of about 4 m in a cooperative swarm using low cost sensors. This is achieved when communication and connectivity in the swarm is reasonably good. Under constrained communication conditions and in the absence of GNSS, the localization accuracy degrades to about 10m. Further it is demonstrated that localization accuracy improves with better network connectivity as well as increase in size of the swarm network.