Generic and modular model to develop virtual laboratories for telerobotics over the internet
Von Borstel Luna, Fernando Daniel
MetadatosMostrar el registro completo del ítem
Sustained advances in information and communication technologies have made it possible to access remote and globally distributed resources for instruction, information and group collaboration over the Internet. In this context, the Virtual Laboratory (VL) concept emerges to offer heterogeneous and distributed environments to perform remote experimentation by operating simulated equipment or remotely operating real equipment. Despite its importance in educational and research fields, VLs are not developed in a systematic way, since its implementation integrates multiple technologies to communicate through the Internet, control heterogeneous equipment, and simulate experiments, among others; VLs are designed and implemented based on the experience of the developer, using intuitive and tested approaches, disregarding what functionalities are needed to perform experimentation, creating partial models that ignore structural composition or behavioral design, using ad hoc architectures or development frameworks that depend on new features of interconnection technologies. Research performed in this work attempts to bridge the gap between intuitive methodologies and a formal and general methodology to develop VLs for telerobotics over the Internet. The performed research creates a generic and modular VL model with three essential elements: Guest, Media, and Host. The proposed model allows the developer to be aware of tested teleoperation strategies. The model is composed of generic entities, which are vendor and technology independent to define the appropriate configuration of desired computer-based processes in a VL for telerobotics. This model is related to the event-based control theory through a teleoperation architecture to introduce control design requirements for stable and synchronized Internet-based VL applications. The model is transformed into a reference framework, which can be customized, based on the experiment specifications. Three functionalities are defined in the reference framework to allow the user to remotely perform a true experiment: mount, define, and execute. The unified modeling language (UML) is used to describe in detail the structure and dynamics of these functionalities based on the object-oriented paradigm. The object-oriented reference framework is taken as the starting point by a general methodology to avoid costly development of new VL applications from scratch. This development methodology has four phases. The first phase analyzes the experiment and defines its functionalities and their inherent components. The second phase identifies and instantiates components into the UML framework. The customized framework describes experiment functionalities as VL subsystems. In the third phase, a novel procedure formally translates the dynamics of the customized framework into the Petri Net notation to carry out a quantitative and qualitative analysis, which relates the Petri Net design with event-based control properties. The fourth phase merges VL subsystems to compose a complete Petri Net design, which is analyzed to validate the sequential execution of VL subsystems. If necessary, the developer can synthesize Petri Net structures to control its behavior. This phase verifies that experiment functionalities perform sequentially, assuring a true experimentation, and validates that the dynamic design represents a stable event-based system over the Internet. The generated structural and dynamic designs provided by UML and Petri Net models supply a guideline to the developer to implement a VL, based on the object-oriented paradigm and the event-based control. Generated UML diagrams allow producing, high quality and low cost software artifacts. Petri Net diagrams provide a comprehensive design of the inherent control of software components. This development methodology proposes an original approach for modeling, designing, analyzing, and validating the structure and dynamics of VL applications for telerobotics. To conclude, two VLs for telerobotics are developed in this study, using the proposed methodology. A VL for mobile robotics is designed and implemented. It uses potential field and computer vision techniques to remotely plan and follow robot trajectories. After that, a VL for bilateral teleoperation that provides visual and haptic feedback is also designed and implemented. It uses a real-time event-based controller to provide sensory information. This VL application reuses software artifacts of the previously designed VL. Experiments are carried out via Internet and Internet2 (Media), using test beds that allow remote interactions between Guest and Host that are geographically separated. All the experimental results confirm the presented theory.