Search

We have developed and demonstrated an intelligent auto-organizing aerial robotic sensor network system composed of cameras installed on UAVs and ground fixtures for urban surveillance using the decentralised control paradigm. The system can auto-organize to meet user requirements, changes to user requirements and when new UAVs or ground cameras are added or are lost due to failure. User designated targets can be automatically tracked and handed over from one camera to another, to ensure continuous tracking. The cameras can also self adjust their positions and pointing angles to cooperatively cover areas of interest completely. We have achieved this by designing a software architecture for managing multiple interacting algorithms online and in parallel, facilitating incremental software development and debugging. We have adapted algorithms to become computationally efficient and effective for real world use online and on-board the UAVs, and ensured that the integration of video analytics and tactical behaviours work in real world settings. We have also achieved near optimal, online close coordination of the UAVs and ground cameras. To the best of our knowledge, this is the first proof-of-concept of a fully integrated intelligent auto-organizing aerial robotic sensor network system of cameras installed on UAVs and ground fixtures for urban surveillance. [ABSTRACT FROM AUTHOR]

Recently, multi-UAV systems are attracting growing interests. It offers wide range of applications in civilian and military environment, carrying out dangerous missions that manned aircraft cannot offer. Currently, an important challenge is the collision avoidance algorithm. The idea is to use the collision avoidance algorithm to control the multi-UAV systems. This will guarantee the safety of the UAVs. The UAVs will complete the missions without colliding with any moving or static obstacles. This topic has motivated the development of various collision avoidance algorithms. In this paper, we proposed some improvements on the three-dimensional velocity obstacle algorithm proposed in Jenie et al. (J. Guid. Control Dyn. 39(10), 2312–2323 25). Our improvements are threefold. First, we indicate the limitations of the original 3D collision avoidance method and present the modifications on the algorithm. Second, we develop the velocity obstacle method to be capable of handling cube obstacles in 3D space. Third, a real flight test is conducted to verify the effectiveness of the proposed three-dimensional velocity obstacle method. [ABSTRACT FROM AUTHOR]

Multi-UAV systems are becoming a true reality. They have wide applications in surveillance, search, hazardous rescue and other civil services. Currently, an important challenge is the robust control against UAV failures. In this paper, the distributed UAV loss detection and auto-replacement scheme is discussed. The basic idea is to use the cooperative method to control the multi-UAV system to accomplish the missions. This is achieved by exchanging heartbeats (HBs) and information fusion. We first use the obtained information to detect if one UAV is lost. Subsequently, an auto-replacement logic is used to send a UAV with low priority to occupy the target position or task which was assigned to the lost UAV. Next, a recovery algorithm is proposed when a newly inserted UAV or a lost UAV is recovered from failures. Finally, the proposed scheme is tested by computer simulations and real experiments. [ABSTRACT FROM AUTHOR]