Ph.D. Dissertation Defense “Adapting the Search Space while Limiting Damage during Learning in a Simulated Flapping Wing Micro Air Vehicle” By Monica Sam

Ph.D. Committee:  Drs. John Gallagher, Advisor, Michael Raymer, Joseph Slater (ME), and Mateen Rizki

ABSTRACT:

Cyber-Physical Systems (CPS) systems characterized by closely coupled physical and software components that operate simultaneously on different spatial and temporal scales; exhibit multiple and distinct behavioral modalities; and interact with one another in ways not entirely predictable at the time of design.  A commonly appearing type of CPS contains one or more "smart components" that adapt locally in response to global measurements of whole system performance.   An example of a smart component robotic CPS system is a Flapping Wing Micro Air Vehicle (FW-MAV) that contains wing motion oscillators that control their wing flapping patterns to enable the whole system to fly precisely after the wings are damaged in unpredictable ways. Localized learning of wing flapping patterns using meta-heuristic search optimizing flight precision has been shown effective in recovering flight precision after wing damage.  However, such methods provide no insight into the nature of the damage that necessitated the learning.  Additionally, if the learning is done while the FW-MAV is in service, it is possible for the search algorithm to actually damage the wings even more due to overly aggressive testing of candidate solutions.  In previous work, a method was developed to extract estimates of wing damage as a side effect of the corrective learning of wing motion patterns.  Although effective, that method lacked in two important respects.  First, it did not settle on wing gait solutions quickly enough for the damage estimates to be created in a time acceptable to a user.  Second, there were no protections against testing too aggressive wing motions that could potentially damage the system even further during the attempted behavior level repair. This work addresses both of those issues by making modifications to the representation and search space of wing motion patterns potentially visited by the online metaheuristic search.  The overarching goals were to lessen the time required to achieve effective repair and damage estimates and to avoid further damage to wings by limiting the search's access to overly aggressive wing motions.  The key challenge was coming to understand how to modify representations and search space to provide the desired benefits without destroying the method's ability to find solutions at all.  With the recent emergence of functional insect-sized and bird-sized FW-MAV and an expected need to modify wing behavior in service, these believed to be the first of their kind studies are particularly relevant.

Selected Publications
Sam, M., Boddhu, S., Gallagher, J.C. (2017).  A dynamic search space approach to improving learning in a simulated flapping wing micro air vehicle,  in the Proceedings of the 2017 IEEE Congress on Evolutionary Computation, IEEE Press,  San Sebastian, Spain. 
Sam, M., Boddhu, S., Duncan, K., Botha, H., Gallagher, J.C. (2016). Improving In-Flight Learning in a Flapping Wing Micro Air Vehicle, International Journal of Monitoring and Surveillance Technologies Research (IJMSTR). Vol. 4., No. 1. 
Gallagher, J.C., Sam, M., Boddhu, S., Matson, E., Greenwood, G. (2016).  Drag force fault extension to evolutionary model consistency checking for a flapping-wing micro air vehicle.  in the 2016 IEEE Congress on Evolutionary Computation 
Sam, M., Boddhu, S., Duncan, K., Gallagher, J.C. (2014). Evolutionary Strategy Approach for Improved In-Flight Control Learning in a Simulated Insect-Scale Flapping-Wing Micro Air Vehicle. in the 2014 IEEE International Conference on Evolvable Systems (ICES) (ICES2014)
Duncan, K., Boddhu, S., Sam, M., Gallagher, J.C. (2014). Islands of Fitness Compact Genetic  Algorithm for Rapid In-Flight Control Learning in a Flapping-Wing Micro Air Vehicle:  A Search Space Reduction Approach. in the 2014 IEEE International Conference on Evolvable Systems (ICES) (ICES2014)