{"607557":{"#nid":"607557","#data":{"type":"event","title":"PhD Defense by Takuma Nakamura","body":[{"value":"\u003Cp\u003EPh.D. Thesis Defense\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eby\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003ETakuma Nakamura\u003C\/strong\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003E(Advisor: Professor Eric N. Johnson)\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003EMultiple-Hypothesis Vision-Based Landing Autonomy\u003C\/strong\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003E1:00 PM, Tuesday, July 24, 2018\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cem\u003EMontgomery Knight 317\u003C\/em\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003EABSTRACT:\u003C\/strong\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003EUnmanned aerial vehicles (UAVs) need humans in the mission loop for many tasks, and\u003C\/p\u003E\r\n\r\n\u003Cp\u003Elanding is one of the tasks that typically involves a human pilot. This is because of the complexity of a\u003C\/p\u003E\r\n\r\n\u003Cp\u003Emaneuver itself and flight-critical factors such as recognition of a landing zone, collision avoidance,\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eassessment of landing sites, and decision to abort the maneuver. Another critical aspect to be\u003C\/p\u003E\r\n\r\n\u003Cp\u003Econsidered is the reliance of UAVs on GPS (global positioning system). A GPS system is not a reliable\u003C\/p\u003E\r\n\r\n\u003Cp\u003Esolution for landing in some scenarios (e.g. delivering a package in an urban city, and a surveillance\u003C\/p\u003E\r\n\r\n\u003Cp\u003EUAV repatriating a home ship with the jammed signals), and a landing solely based on a GPS\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eextremely decreases the UAV operation envelope. Vision is promising to achieve fully autonomous\u003C\/p\u003E\r\n\r\n\u003Cp\u003Elanding because it is a rich-sensing, light, affordable device that functions without any external\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eresource. Although vision is a powerful tool for autonomous landing, the use of vision for state\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eestimation requires extensive consideration. Firstly, vision-based landing faces a problem of occlusion.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EThe target detected at a high altitude would be lost at certain altitudes while a vehicle descends;\u003C\/p\u003E\r\n\r\n\u003Cp\u003Ehowever, a small visual target can not be recognized at high altitude. Second, standard filtering\u003C\/p\u003E\r\n\r\n\u003Cp\u003Emethods such as extended Kalman filter (EKF) faces difficulty due to the complex dynamics of the\u003C\/p\u003E\r\n\r\n\u003Cp\u003Emeasurement error created due to the discrete pixel space, conversion from the pixel to physical units,\u003C\/p\u003E\r\n\r\n\u003Cp\u003Ethe complex camera model, and complexity of detection algorithms. The vision sensor produces an\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eunfixed number of measurements with each image, and the measurements may include false positives.\u003C\/p\u003E\r\n\r\n\u003Cp\u003EPlus, the estimation system is excessively tasked in realistic conditions. The landing site would be\u003C\/p\u003E\r\n\r\n\u003Cp\u003Emoving, tilted, or close to an obstacle. The available landing location may not be limited to one. In\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eaddition to assessing these statuses, understanding the confidence of the estimations is also the tasks of\u003C\/p\u003E\r\n\r\n\u003Cp\u003Ethe vision, and the decisions to initiate, continue, and abort the mission are made based on the\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eestimated states and confidence. The system that handles those issues and consistently produces the\u003C\/p\u003E\r\n\r\n\u003Cp\u003Enavigation solution while a vehicle lands eliminates one of the limitations of the autonomous UAV\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eoperation. This thesis presents a novel state estimation system for UAV landing. In this system, vision\u003C\/p\u003E\r\n\r\n\u003Cp\u003Edata is used to both estimate the state of the vehicle and map the state of the landing target (position,\u003C\/p\u003E\r\n\r\n\u003Cp\u003Evelocity, and attitude) within the framework of simultaneous localization and mapping (SLAM). Using\u003C\/p\u003E\r\n\r\n\u003Cp\u003Ethe SLAM framework, the system becomes resilient to a loss of GPS and other sensor failures. A novel\u003C\/p\u003E\r\n\r\n\u003Cp\u003Evision algorithm that detects a portion of the marker is developed, and the stochastic properties of the\u003C\/p\u003E\r\n\r\n\u003Cp\u003Ealgorithm are studied. This algorithm extends the detectable range of the vision system for any known\u003C\/p\u003E\r\n\r\n\u003Cp\u003Emarker. However, this vision algorithm produces highly nonlinear, non-Gaussian, and multi-modal\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eerror distribution, and a naive implementation of filters would not accurately estimate the states. A\u003C\/p\u003E\r\n\r\n\u003Cp\u003Evision-aided navigation algorithm is derived within extended Kalman particle filter (PF-EKF) and Rao-\u003C\/p\u003E\r\n\r\n\u003Cp\u003EBlackwellized particle filter (RBPF) frameworks in addition to a standard EKF framework. These\u003C\/p\u003E\r\n\r\n\u003Cp\u003Emulti-hypothesis approaches not only deal well with a highly non-linear and non-Gaussian distribution\u003C\/p\u003E\r\n\r\n\u003Cp\u003Eof the measurement errors of vision but also results in numerically stable filters. The computational\u003C\/p\u003E\r\n\r\n\u003Cp\u003Ecosts are reduced compared to a naive implementation of particle filter, and these algorithms run in real\u003C\/p\u003E\r\n\r\n\u003Cp\u003Etime. This system is validated through numerical simulation, image-in-the-loop simulation, and flight\u003C\/p\u003E\r\n\r\n\u003Cp\u003Etests.\u003C\/p\u003E\r\n\r\n\u003Cp\u003E\u003Cstrong\u003ECOMMITTEE MEMBERS:\u003C\/strong\u003E\u003C\/p\u003E\r\n\r\n\u003Cp\u003EProfessor Eric N. Johnson, School of Aerospace Engineering (Advisor)\u003C\/p\u003E\r\n\r\n\u003Cp\u003EProfessor Panagiotis Tsiotras, School of Aerospace Engineering\u003C\/p\u003E\r\n\r\n\u003Cp\u003EProfessor Eric Feron, School of Aerospace Engineering\u003C\/p\u003E\r\n\r\n\u003Cp\u003EProfessor James Hays, School of Computer Science\u003C\/p\u003E\r\n\r\n\u003Cp\u003EProfessor Patricio Antonio Vela, School of Electrical Engineering\u003C\/p\u003E\r\n","summary":null,"format":"limited_html"}],"field_subtitle":"","field_summary":"","field_summary_sentence":[{"value":"Multiple-Hypothesis Vision-Based Landing Autonomy"}],"uid":"27707","created_gmt":"2018-07-06 18:54:14","changed_gmt":"2018-07-06 18:54:14","author":"Tatianna Richardson","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2018-07-24T14:00:00-04:00","event_time_end":"2018-07-24T16:00:00-04:00","event_time_end_last":"2018-07-24T16:00:00-04:00","gmt_time_start":"2018-07-24 18:00:00","gmt_time_end":"2018-07-24 20:00:00","gmt_time_end_last":"2018-07-24 20:00:00","rrule":null,"timezone":"America\/New_York"},"extras":[],"groups":[{"id":"221981","name":"Graduate Studies"}],"categories":[],"keywords":[{"id":"100811","name":"Phd Defense"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[{"id":"1788","name":"Other\/Miscellaneous"}],"invited_audience":[{"id":"78761","name":"Faculty\/Staff"},{"id":"177814","name":"Postdoc"},{"id":"78771","name":"Public"},{"id":"174045","name":"Graduate students"},{"id":"78751","name":"Undergraduate students"}],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}