| Over the past decade, there has been substantial research conducted in multi-robot development techniques. However, this advancement still cannot cope with the ever increasing scale and complexity of building autonomous swarm-robotic systems to perform complex tasks in open environments. One approach for overcoming these challenges is to develop advanced multi-robot control architectures to increase the overall system coordination, achieve well-organized behavior, and improve system dependability.;In this dissertation, we present an advanced virtual device based tele-control framework in which sliding autonomy, distributed AI planning, behavior-based self-stabilization, and high dependability assurance techniques are seamlessly integrated together. This framework facilitates the operator to rapidly deploy and control a group of robots to perform sophisticated tasks through a system supported group behavior set defined by the virtual device.;In the advanced virtual device control framework, the fundamental structure formation capability is incorporated into the system based on the self-stabilization principle. Existing self-organizing or self-assembly methods are either not very powerful or are not very efficient or scalable. This dissertation presents efficient algorithms to enable a robotic swarm system to form various structures even though each individual mobile robot in the system is equipped with only limited sensors. Moreover, the inherent characteristic of the self-stabilization algorithm enables the structure formed by the robots to be robust, reliable, and capable of tolerating the failure of individual robots.;The high level structure primitive, as well as real-time functionality, are supported by hierarchical planners in our framework. The Dissertation presents centralized and decentralized path planners and motion planners to dynamically optimize the global plan based on the individual robot's local information. User commands can be automatically decomposed, distributed and executed on the group of robots.;Another important feature of our architecture is that it supports offline/online dependability analysis. We have developed various models and analysis techniques to improve the reliability of the multi-robot systems, enabling the system to continuously perform complex tasks even in extremely hazardous environments.;This research has a significant impact on multi-robot control architectures. With the novel flexible virtual device, each individual robot has its own autonomy to adapt to environmental changes or internal failures, while still collectively providing a uniform service to the operator. Thereby, it increases system reliability, enhances system capability, and improves user-friendliness.;The results of the experimental evaluations confirm that the virtual device framework can be effectively applied to various rigid-body transportation applications. |
