@phdthesis{,
    author = {Strenzke, Ruben},
    title = {Cooperation of Human and Artificial Intelligence on the Planning and Execution of Manned-Unmanned Teaming Missions in the Military Helicopter Domain : Concept, Requirements, Design, Validation},
    editor = {},
    booktitle = {},
    series = {},
    journal = {},
    address = {},
    publisher = {},
    edition = {},
    year = {2019},
    isbn = {},
    volume = {},
    number =  {},
    pages = {},
    url = {},
    doi = {},
    keywords = {Manned-unmanned Teaming, Human-Machine Collaboration, Mixed-Initiative Planning, Assistant System, UAV operator, Mission Planning, Automated Planning, PDDL},
    abstract = {This dissertation provides arguments in favor of the cooperation of human operators and an Artificial Intelligence (A.I.) based software on the in-flight planning and the execution of joint manned and unmanned military helicopter missions. Furthermore, it presents requirements concerning the design and performance of such an A.I. based cooperative software. In order to build the argumentation, a brief insight on the different cognitive advantages of the human and A.I. in general is given. After deriving a concept of human-automation cooperation best described as “mixed initiative” from this and describing the design and implementation of a cooperative A.I. system prototype, an empirical analysis with subject matter experts (SMEs) as test persons provides further arguments in favor of this approach. The cooperation concept presented here is based on the assumption that it is hardly possible to know in advance, which subtask (or subproblem) is suited to be allocated to either of the two partners, human or automation. This is because both have different abilities, and each task or problem can have a different structure demanding specific problem-solving abilities. Hence, each of the partners in the mixed human-machine team should be able to take problem-solving initiative, i.e. the initiative for adapting or evolving the mission plan and start task execution. This constitutes the mixed-initiative planning, replanning and plan execution approach, which is the basis of the Mixed-Initiative Mission Planner (MMP) that has been designed and implemented by the author in a way that makes use of symbolical A.I. technology (classical, domain-independent Automated Planning). The empirical analysis has been performed as full-mission simulation experiments with professional German Army Aviators as SME test persons in two main configurations: with support by the A.I. mission planning and execution component and without this support. As this dissertation shows, the analysis has been validated by the SMEs to be representative. However, due to the small pool of eight test persons that makes statistical validation difficult, the main evidence is based on their systematically collected subjective feedback showing that the advices and solutions given by the MMP were considered rather useful than hindering. Finally, this dissertation is able to provide a basis for the requirements and the design of cooperative A.I. systems in this field of application. This is achieved also by the systematic collection of subjective feedback concerning user requirements from the test persons after the experimental campaign. The feedback also includes information on alternative mission-planning work-flows and human-machine interface designs.},
    note = {},
    
    school = {Universität der Bundeswehr München},
    
}