With a problem such as search and rescue where human lives are at risk, the time taken to locate and subsequently rescue a victim is the key factor and is often limited by cost and the physical resources available to the search team. This thesis is concerned with locating lost targets, mainly human victims in a potentially large poorly defined area in a time optimal manner. This problem is important within the field of search and rescue, where the recent advancement in unmanned technologies can be used to boost the searching capability. By utilising unmanned aerial systems to increase the number of search vehicles involved in the effort, there exists the ability for a significant increase in search and rescue capability. However, coordinating multiple autonomous vehicles in a time optimal manner from a remote ground station presents many challenges. Optimal motion planning and the ability to effectively monitor and control a possible swarm of unmanned vehicles are critical components.In a typical Unmanned Aerial System (UAS) application, the flight path is generally decided upon by a human controller using stochastic methods with the full mission plan programmed in prior to launch. This is usually done through the use of pre-determined loiter or search patterns of or by manually plotting a series of GPS way points for the flight management system to follow. The work contained within this thesis presents a strategy of using stochastic methods to estimate the location of the lost target or targets. This non-deterministic approach when coupled with multiple computer simulation techniques creates a novel control environment. A decentralised approach is presented that allows for maximum scalability and robustness.The developed simulator at the University of New South Wales was built to demonstrate and study the optimisation of a search and rescue mission based on the last known location of a lost target. The presented simulation environment adds significant situational awareness to the otherwise limited milieu. The proposed framework incorporates a simulation environment that is capable of interfacing with real search vehicles to create not just a human in the loop environment but also a system-in-the-loop environment where the complete developed system is an integral part of the cooperative control and simulation ground station utilising all assets manned, autonomous and remotely controlled.
展开▼