Space-based Task Allocation in Robotic Fleets with Adaptive Autonomy

Networked robotic fleet consists of multiple robots with heterogeneous capabilities that jointly perform a mission. Due to the ability to distribute heterogeneous robots in a working environment and exploit their cooperation and collaboration capacities, it has been shown that utilizing a robotic fleet for accomplishing a given mission is faster and more efficient than using a single robot. Nowadays, robotic fleets are mostly limited to execute missions in structured and controlled environments where a centralized system manages and supervises mission execution. With the goal to shift a focus from centralized to distributed control systems, this thesis explores the potential to use robotic fleets in heterogeneous and unstructured environments in scenarios requiring collaboration in mixed human-robot teams. It addresses new challenges related to communication, task allocation, coordination, collaboration, cooperation, and adaptive autonomy, and compares the well-established centralized coordination approach with several new distributed approaches. Reliable communication between distributed heterogeneous team members is a necessary condition for efficient coordination. An efficient coordination model is a critical enabler for task allocation which is a fundamental problem in multi-robot systems where the core requirement is to find an optimal set of heterogeneous robots that have to cooperate to execute a complex mission. To work together on complex tasks, humans and robots have to conform to efficient collaboration and cooperation models. The transition to cooperation and collaboration mode implies the change of role a human has as a remote operator in traditional systems to the peer who jointly performs tasks with the fleet members. The thesis focuses on a development of a general coordination framework with a distributed decision making system deployed on each robot in a fleet. The coordination framework is based on the Space-Based Computing (SBC) paradigm extended with the Semantic Web Technologies resulting in semantic shared data space which allows annotating shared data with machineunderstandable metadata. While the SBC paradigm provides for strongly decoupled architecture and preserves autonomy of the interacting team members, processing the metadata introduces flexibility in modelling collaboration processes in the fleet. For the evaluation of the developed coordination framework, established is a comprehensive benchmark setting utilized to quantify concrete performance indicators of different coordination approaches. Moreover, it is analyzed which scenarios are best suited for the developed coordination approaches. The results indicate that adaptive autonomy and shared knowledge improve performance in task allocation in complex missions.