Cooperative Adaptive Control for Multi-Agent Systems

The robotics system have extensive applications in various fields, such as underwater environment survey, detecting hazardous materials, space exploration, and robotic manufacturing. These scenarios can benefit from the use of multi robot systems where the tasks are difficult to be accomplished by an individual robot. Inspired by nature and human society, a team of robots usually provides redundancy and reliability, and cooperatively finished the complicated tasks. This dissertation explores a novel cooperative adaptive theory which enables robots to collaborate resiliently in a highly uncertain environment by fully distributed control framework. A distributed cooperative adaptive control framework is proposed that utilizing fast adaptation to estimates input-output equivalent time-varying uncertainties not only from environments but also from control decisions of neighboring robots. This framework further individually generates collaborative control signal based on the estimation results. One hypothesis is that the fast adaptation in the estimation and low-pass filtering mechanism in control signal generation will break the algebraic loops between robots and stabilize the entire network. As a result, each individual robot only contributes what is needed in a collaborative mission based on real time evaluation of efforts from other robots. This leads to a more resilient and flexible collaborative robotic network especially during unexpected situations. Intensive mathematical analysis for different scenarios, such as linear, nonlinear, and heterogeneous multi-agent systems, are given to demonstrate how the local adaptation can lead to global stability. It also enhances human’s understanding of the collective behavior in nature. Finally, several typical collaborative robot network applications, such as formation flying and industrial robot synchronization, be applied to demonstrate the proposed cooperative adaptive control framework