UAS Contingency Management Autonomy with Experimentally Validated Models
dc.contributor.author | Sharma, Prashin Santosh | |
dc.date.accessioned | 2022-09-06T16:06:16Z | |
dc.date.available | 2022-09-06T16:06:16Z | |
dc.date.issued | 2022 | |
dc.date.submitted | 2022 | |
dc.identifier.uri | https://hdl.handle.net/2027.42/174332 | |
dc.description.abstract | Uncrewed Aerial Systems (UAS) are increasingly deployed for surveillance and transport applications. However, their safety and performance are significant concerns. This dissertation develops a risk-aware autonomy architecture backed by experimentally validated performance and degradation models necessary to maintain acceptable risk levels even when flying at low altitude in urban areas. The first contribution is an experimental model of aerodynamic performance for tractor and pusher hexacopter configurations. This work was motivated by experimental analysis of a single propulsion unit with different propeller configurations for which a pusher configuration generated 20% more thrust than tractor configurations. Wind tunnel experiments yielded the insight that in static conditions, the pusher hexacopter has a higher lift-to-weight ratio than the tractor configuration because the pusher generates ~15% more thrust than the tractor. However, in forward flight this higher lift-to-weight ratio is traded against a lower lift-to-drag ratio for the pusher design that has ~25% more drag than the tractor design. We verified these results by conducting outdoor autonomous flight tests. These results further motivated an investigation of wind sensing sensitivity for a hexacopter in tractor and pusher configurations. Wind sensing experiments and analysis revealed that the pusher hexacopter configuration offers higher sensitivity to wind fluctuations than the tractor hexacopter. The second contribution is a battery and motor reconfiguration scheme in a multi-battery pack to assure a UAS has sufficient stored energy to reach its destination. The proposed reconfiguration scheme is proactive by design, utilizing component failure predictions from model-based prognostic methods. A model for the LiPo battery is experimentally determined using a novel low-cost testbed to collect charge/discharge data for battery model parameter identification. Fault modes of BLDC motors are studied, and a technique for motor fault prognosis is presented. Battery and motor degradation models are used for prognosis, providing End of Discharge and Remaining Useful Life estimates, respectively. Using abstractions of EOD value and other critical state features, a novel battery reconfiguration MDP is proposed for a series-parallel battery pack. The MDP policy optimally reconfigures the battery pack in flight. Case studies are presented to demonstrate benefits of the battery reconfiguration MDP. The third contribution is a MDP-based Contingency Management Autonomy (CMA) capability to generate mission-level directives that preserve safety when component reconfiguration alone is insufficient. Although component reconfiguration prevents most sudden mission failures, there is no guarantee a degraded UAS can safely complete its planned flight. Optimal CMA policy effectiveness is evaluated on a high-fidelity simulator using experimentally validated models. Metrics such as mission failure rate are used to analyze CMA MDP performance over $900$ Monte Carlo simulations. In poor battery health conditions, CMA MDP policy has a failure rate of 1.3% compared to a baseline policy’s failure rate of 71%. In summary, this thesis contributes to better understanding multicopter flight performance and to improving safety of small UAS flight. Safety is addressed using an abstract MDP decision-making approach due to vehicle and operational complexity. Component reconfiguration and contingency management will be instrumental in deploying autonomous UAS. This dissertation provides a baseline capability on which future component and systems performance and prognostics elements can be added. | |
dc.language.iso | en_US | |
dc.subject | UAS, contingency management, aerodynamics, wind sensing, Markov Decision Process, Reinforcement Learning, Lipo Batteries | |
dc.title | UAS Contingency Management Autonomy with Experimentally Validated Models | |
dc.type | Thesis | |
dc.description.thesisdegreename | PhD | en_US |
dc.description.thesisdegreediscipline | Robotics | |
dc.description.thesisdegreegrantor | University of Michigan, Horace H. Rackham School of Graduate Studies | |
dc.contributor.committeemember | Atkins, Ella Marie | |
dc.contributor.committeemember | Seiler, Peter Joseph | |
dc.contributor.committeemember | Jeannin, Jean-Baptiste | |
dc.contributor.committeemember | Sunberg, Zachary Nolan | |
dc.contributor.committeemember | Woolsey, Craig A | |
dc.subject.hlbsecondlevel | Aerospace Engineering | |
dc.subject.hlbsecondlevel | Computer Science | |
dc.subject.hlbsecondlevel | Electrical Engineering | |
dc.subject.hlbsecondlevel | Transportation | |
dc.subject.hlbtoplevel | Engineering | |
dc.description.bitstreamurl | http://deepblue.lib.umich.edu/bitstream/2027.42/174332/1/prashinr_1.pdf | |
dc.identifier.doi | https://dx.doi.org/10.7302/6063 | |
dc.identifier.orcid | 0000-0003-2828-4087 | |
dc.identifier.name-orcid | Sharma, Prashin; 0000-0003-2828-4087 | en_US |
dc.working.doi | 10.7302/6063 | en |
dc.owningcollname | Dissertations and Theses (Ph.D. and Master's) |
Files in this item
Remediation of Harmful Language
The University of Michigan Library aims to describe library materials in a way that respects the people and communities who create, use, and are represented in our collections. Report harmful or offensive language in catalog records, finding aids, or elsewhere in our collections anonymously through our metadata feedback form. More information at Remediation of Harmful Language.
Accessibility
If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.