This thesis presents an architecture for task independent autonomous flying vehicles. We review the conventional architectures such as the model-and-planner, behavior-based, and the subsumption architectures. We focus on their mechanism, their basic assumptions and their weaknesses. Based on the review, we propose the Strategic Architecture, an adaptive composition of a set of strategic multi-agents, which over-comes limitations of the existing architectures. In the Strategic Architecture, each strategy is designated to deal with a particular task and the composition of strategies tackles more complex tasks. The proposed Strategic Architecture maintains a uniform design for its low level control, high level planning, and multi-robot cooperative system thus facilitating the strategies' c...[ Read more ]

This thesis presents an architecture for task independent autonomous flying vehicles. We review the conventional architectures such as the model-and-planner, behavior-based, and the subsumption architectures. We focus on their mechanism, their basic assumptions and their weaknesses. Based on the review, we propose the Strategic Architecture, an adaptive composition of a set of strategic multi-agents, which over-comes limitations of the existing architectures. In the Strategic Architecture, each strategy is designated to deal with a particular task and the composition of strategies tackles more complex tasks. The proposed Strategic Architecture maintains a uniform design for its low level control, high level planning, and multi-robot cooperative system thus facilitating the strategies' composition. This highly modular organization reduces the dependency between strategies and, hence, increases the reusability of strategies.

Based on the Strategic Architecture, we have designed an autonomous system for model helicopter that takes the advantages of this new architecture. The design consists of well-organized layers of strategies tackling different subtasks using different algorithms, such as the two-dimensional land covering algorithm. The design, on the other hand, serves as an example that substantiates the concepts in the Strategic Architecture.

This thesis has four components: the Strategic Architecture, an autonomous helicopter design, a 2-dimensional land covering algorithm and the simulation. The Strategic Architecture is proposed to overcome the limitations and encompass the strengths of the existing architectures for autonomous system. The autonomous helicopter design shows how different strategies using different algorithms fit into the homogenous Strategic Architecture. Using the two-dimensional land covering task as an example, the autonomous model helicopter system is demonstrated to be applicable in the simulation. The simulation system examines the autonomous system before the autonomous system is built on the actual hardware. The contributions of this thesis are, first, a new Strategic Architecture with strengths in adaptive composition, hardware independence and with a highly reusable module design of strategies and, second, simulation that adequately substantiates these benefits. This research provides strong support for further exploration about automating real helicopters.