Towards concurrent sensor & path planning - a survey of planning methods applicable to UAV surveillance

Autonomous path and sensor planning of a UAV with EO/IR sensors for surveillance and exploration is a very challenging problem. Realistic models of environment, sensors and platforms are very complex, due to the nonlinear and stochastic properties of the world. Hence, algorithms and methods solving realistic planning problems are computationally very demanding. To develop a monolithic planner is probably an impossible task; instead a hierarchical decomposition is necessary. The issue then, is how to decompose the problem into sub-problems that guarantee that the overall objective is achieved. The purpose of this survey is to present suitable methods and approaches that can be applied to surveillance and exploration with autonomous UAVs; in particular, concurrent sensor and path planning methods. First the UAV surveillance and reconnaissance problem is introduced. An overview of research fields and communities related to path planning and/or sensor planning is then given. The main parts of this report are presentations of methods, techniques and approaches to path planning and sensor planning. Furthermore, a brief survey of related research in Sweden is given. The concurrent path and sensor planning problem is very complex and no method has been found that solves the problem in a satisfying way. However, some promising directions are identified and since much research related to this area is in progress, interesting results will hopefully appear during the next few years.