Modified model free dynamic programming :an augmented approach for unmanned aerial vehicle.

The design complexities of trending UAVs necessitates formulation of C ontrol L aws that are both robust and model-free besides being self-capable of handling the evolving dynamic environments. In this research, a unique intelligent control architecture is presented which aims at maximizing the glide range of an experimental UAV having unconventional controls. To handle control complexities, while keeping them computationally acceptable, a distinct RL technique namely Modified Model Free Dynamic Programming (MMDP) is proposed. The methodology is novel as RL based Dynamic Programming algorithm has been specifically modified to configure the problem in continuous state and control space domains without knowledge of the underline UAV model dynamics. Major challenge during the research was the development of a suitable reward function which helps in achieving the desired objective of maximising the glide performance. The efficacy of the results and performance characteristics, demonstrated the ability of the presented algorithm to dynamically adapt to the changing environment, thereby making it suitable for UAV applications. Non-linear simulations performed under different environmental and varying initial conditions demonstrated the effectiveness of the proposed methodology over the conventional classical approaches. [ABSTRACT FROM AUTHOR]