In this paper, we present a model-free experimental method to find a control strategy for achieving stable and autonomous, from a control perspective, flight of a dual-actuator biologically inspired flapping-wing flying microrobot. The main idea proposed in this work is the sequential tuning of parameters for an increasingly more complex strategy in order to sequentially accomplish more complex tasks: upright stable flight, straight vertical flight, and stable hovering with altitude and position control. Each term of the resulting multiple-inputmultiple-output (MIMO) controller has a physical intuitive meaning and the control structure is relatively simple, such that, it could potentially be applied to other kinds of flapping-wing robots.
I. INTRODUCTIONExperiments demonstrating the first controlled vertical unconstrained flight of a 83-mg flapping-wing flying microrobot were presented in [1]. There, the idea of using separate actuators exclusively for control was introduced and demonstrated, through static and flying experiments. The argument for designing, developing, and integrating separate actuators exclusively for control is biologically inspired, based on evidence suggesting that insects in nature employ separate muscles for power and control, respectively [2]. There are important practical problems that arise in the fabrication process developed for materializing the design in [1]. Specifically, fabrication needs to be essentially perfect in order to avoid asymmetries in the prototypes that would make them very difficult to stabilize and control, or uncontrollable. Despite these fabrication challenges, unconstrained flight control of the prototype in [1] was preliminarily, but convincingly, demonstrated using the ideas and findings on altitude control and pitch control in [3]-[5], and references therein.A different design approach, first proposed in [6], was fully developed in the work presented in [7], which is the basic design of the robotic prototype we consider in this paper. This design consists of dual completely independent power actuators that drive each of the wings independently through two separate transmissions, and departs significantly from the previous purely biologically inspired robotic models in [1]. An adaptive model-based control strategy for the prototype in [7] was proposed and tested in the work presented in [8]. In this paper, we propose a new control strategy, which is entirely experimental and model-free, but takes advantage of the knowledge on flapping-wing systems gathered through static and flying experiments in [3]- [5].In this paper, we provide evidence that the control philosophy first proposed in [1], based on asymmetrical flapping patterns, is applicable to the general flapping-wing flight control case. The experimental results presented here are significantly better than those presented in [1], mainly because the design considered in this work (in Figs. 1 and 2) is, from a practical perspective, more controllable and robust to fabrication errors. The main new idea explor...