Multi-Object Tracking Techniques Using Mobile Sensor Networks

Recently,Mobile Sensor Networks(MSNs)have been widly used for robotic searchand-rescue,border patrols,battle scenarios and environmental monitoring.An example can be a group of Unmanned Aerial Vehicles(UAVs),typically organized into swarm using local area networks to achieve real-time monitoring of the Area of Interest(Ao I).Due to the limited Field of View(Fo V)of the sensor nodes,the decentralized structure of the network,the potentially high mobility of the UAVs,and the high accuracy and robustness required for a fully autonomous operation,multi-object searching and tracking over Ao I are difficult tasks.This thesis focus on the multi-object tracking algorithm for MSNs that are capable of sensing,processing,mobilization and communication with other nodes.In order to solve the multi-object searching and tracking problem,there are the following three basic difficulties:1.how to optimally deploy a set of mobile sensors in a given AOI in such a way that maximizes the area coverage?2.how to handle multiple-object tracking problems considering a varying and unknown number of objects with unknown motion models?3.How to build an effective distributed(decentralized)framework to achieve joint searching and tracking of regional multi-maneuvering targets?Compared with previous research on regional monitoring of MSNs,the main contents of this thesis are as follows:1.Inspired by the solitary behavior of some animals,the anti-flocking control model is used to maximize the area coverage,which enables the distributed collaborative searching for unknown area.The described anti-flocking strategy exhibits exhibits prominent performance in terms of the area coverage and the detection effectiveness;2.In order to meet the need of perception and exploratio of the unknown environment for UAVs swarm,a multi-UAV collaborative searching and intelligent projectile system is developed.The system can fully simulate the process of simulating multiUAV searching and strike;3.This thesis extends the multi-target filtering algorithm based on the concept of ran-dom finite set,to handle searching and tracking problems with limited Fo V sensors;4.The proposed framework for joint searching and tracking of multi-maneuvering objects can handle a varying and unknown number of objects with unknown motion models.It can not only estimate the state of multiple objects,but also support the track management.Finally,the simulation results verify the effectiveness of the Anti-Flocking distributed collaborative control algorithm and framework for joint searching and tracking of multimaneuvering objects.In addition,the flight tests demonstrate the effectiveness and robustness of the swarm searching system.