We proposed an RL framework for control of flapping-wing MAVs (2006). The algorithm has been discussed and simulation results using a quasi-steady model showed initial promise. In this paper, the results from an experiment on a Drosophila-based dynamically scaled model are presented and are used to verify the control framework. Moreover, a comparison between a biological Drosophila melanogaster and the experimental results shows the actual possibility of employing the proposed approach to MAV control problem

Flapping micro aerial vehicles are interesting in applications where maneuverability is needed in confined spaces. Yet aerodynamics of insect flapping flight is not completely known and the research in this area lacks a suitable aerodynamic model that can be used for control purposes. Reinforcement learning approach is proposed which is inspired from "how" the learning is achieved in real insects in nature. The reinforcement learning controller has been simulated using a quasi-steady aerodynamic model and shown to converge to a flapping motion. The learning capacity and advantages of this approach are also discussed

Insect sensors are fascinating for their simplicity in design, efficiency in function, and robustness to variable environments. Many engineering applications would benefit from these features, therefore, much research, especially recently in biorobotics, has focused on biomimetic sensors. However, human engineering ingenuity still falls short in comparison to nature's capabilities. This research area is inherently multidisciplinary in nature, calling for collaboration from biologists, neurologists and engineers. This paper is intended to provide an insight into force sensor design for microflight based on current biological, biomimetic and miniature sensors available to compare existing attempts and to make a framework to identify issues that need further attention. This review covers recent experiments concerning the sensor design, as well as different key works in this area. The limitations and scaling issues with some of these methods are explored.