Combining Behavior Trees and State Machines for Mission Planning in Autonomous Flight

This bachelor’s thesis explores a novel approach to mission planning in autonomous flight. In the ever-evolving field of autonomous aerial vehicles, efficient and adaptable mission planning is crucial for the safe and effective operation of the aerial system.
Traditionally, mission planning has relied on many methods including behavior trees and state machines to guide the decision-making process of autonomous drones. Behavior trees offer flexibility in representing complex behaviors but can be challenging to manage when multiple behaviors interact. On the other hand, state machines provide clear transitions between states, simplifying control logic, but may struggle to handle intricate and dynamic missions systematically.
This research proposes a three-layer system, the top layer being a behavior tree responsible for following the overall mission objectives and higher-level decision-making. The second layer is a state machine responsible for ensuring the system’s safety (for example, ensuring that the emergency state can always be reached). The last layer involves individual behavior trees for each of the states. These layers are meant to interact with each other, providing vital information such as the current state and state transitions to be executed. This system intends to create a more versatile and robust mission planning system. By integrating the two concepts, this thesis aims to combine the benefits of each approach while trying to overcome their individual limitations. The resulting hybrid framework intends to offer increased adaptability, reliable decision-making, and enhanced autonomy during flight.
With continuous growth in the field of autonomous aviation, the outcomes of this research aim to contribute meaningfully to the decision-making algorithms for use in autonomous systems, enhancing their versatility and adaptability in various scenarios. By combining existing technologies, this study seeks to pave the way for more efficient and reactive autonomous aerial operations, with applications extending to various domains, including surveillance, delivery, and transportation.