Biblio

Insect-Scale Flapping-Wing Micro-Air Vehicles (FW-MAVs), can be particularly sensitive to control deficits caused by ongoing wing damage and degradation. Since any such degradation could occur during flight and likely in ways difficult to predict apriori, any automated methods to apply correction would also need to be applied in-flight. Previous work has demonstrated effective recovery of correct flight behavior via online (in service) evolutionary algorithm based learning of new wing-level oscillation patterns. In those works, Evolutionary Algorithms (EAs) were used to continuously adapt wing motion patterns to restore the force generation expected by the flight controller. Due to the requirements for online learning and fast recovery of correct flight behavior, the choice of EA is critical. The work described in this paper replaces previously used oscillator learning algorithms with an Evolution Strategy (ES), an EA variant never previously tested for this application. This paper will demonstrate that this approach is both more effective and faster than previously employed methods. The paper will conclude with a discussion of future applications of the technique within this problem domain.

Previously, we introduced Evolutionary Model Consistency Checking (EMCC) as an adjunct to Evolvable and Adaptive Hardware (EAH) methods. The core idea was to dual-purpose objective function evaluations to simultaneously enable EA search of hardware configurations while simultaneously enabling a model-based inference of the nature of the damage that necessitated the hardware adaptation. We demonstrated the efficacy of this method by modifying a pair of EAH oscillators inside a simulated Flapping-Wing Micro Air Vehicle (FW-MAV). In that work, we were able to show that one could, while online in normal service, evolve wing gait patterns that corrected altitude control errors cause by mechanical wing damage while simultaneously determining, with high precision, what the wing lift force deficits that necessitated the adaptation. In this work, we extend the method to be able to also determine wing drag force deficits. Further, we infer the now extended set of four unknown damage estimates without substantially increasing the number of objective function evaluations required. In this paper we will provide the outlines of a formal derivation of the new inference method plus experimental validation of efficacy. The paper will conclude with commentary on several practical issues, including better containment of estimation error by introducing more in-flight learning trials and why one might argue that these techniques could eventually be used on a true free-flying flapping wing vehicle.

Those employing Evolutionary Algorithms (EA) are constantly challenged to engineer candidate solution representations that balance expressive power (I.E. can a wide variety of potentially useful solutions be represented?) and meta-heuristic search support (I.E. does the representation support fast acquisition and subsequent fine-tuning of adequate solution candidates). In previous work with a simulated insect-like Flapping-Wing Micro Air Vehicle (FW-MAV), an evolutionary algorithm was employed to blend descriptions of wing flapping patterns to restore correct flight behavior after physical damage to one or both of the wings. Some preliminary work had been done to reduce the overall size of the search space as a means of improving time required to acquire a solution. This of course would likely sacrifice breadth of solutions types and potential expressive power of the representation. In this work, we focus on methods to improve performance by augmenting EA search to dynamically restrict and open access to the whole space to improve solution acquisition time without sacrificing expressive power of the representation. This paper will describe some potential restriction/access control methods and provide preliminary experimental results on the efficacy of these methods in the context of adapting FW-MAV wing gaits.