Practical motion planning for aerial-like virtual agents in Meta!Blast: A full and complex three dimensional virtual environment

    Campbell, Princess
    Major Professor
    Eve Wurtele
    Julie Dickerson
    Committee Member
    Journal Title
    Journal ISSN
    Volume Title
    Research Projects
    Organizational Units
    Journal Issue
    Electrical and Computer Engineering

    Motion planning, or enabling agents to navigate around a virtual environment autonomously, is an essential requirement for video games and simulations. A well implemented motion planning technique can create a realistic and immersive user experience. If motion planning is not implemented properly, agents will exhibit unrealistic behavior and cause a distraction for the user. Motion planning is often difficult to implement due to the agents' movement capabilities and the complexity of the virtual environment in which the agents exist. In a traditional three dimensional video game in which the agents are bound by gravity, the agents' motion takes place mostly in the XZ-plane. In other words, the agents' degree of freedom (DOF) is three. In this case, motion planning is translated into a two-dimensional problem, which is relatively easier to compute. However, when the agents can move in any three dimensional direction or to any three dimensional position in space, motion planning is much more complex.

    Meta!Blast is a three dimensional educational video game. Implementing motion planning in Meta!Blast is challenging for three reasons: The first reason is the agents have at least six degrees of freedom and can be translated or rotated about any axis in the three dimensional virtual environment. The second reason is the agents exist in a dense environment with many irregularly shaped models that need to be considered during planning. Lastly, Meta!Blast will be deployed in the high school classroom where computer hardware resources are limited, eliminating some planning techniques found in the literature. This thesis provides a practical solution for high DOF agents in dense environments using a combination of octree space partitioning, A* path-planning, and steering behaviors.