PEARL : A Generic Architecture for Live Experiments in aRemote

The convenience and widened access ooered by computer based learning is already established. However, this pedagogy can be enhanced by the opportunity for students to plan and perform "live" experiments over the Internet (as opposed to simulations). This paper describes the PEARL generic architecture for remote experiments and our initial experience implementing computer vision experiments using it. In developing the PEARL architecture, we have tried to accommodate a wide range of (low-level) equipment interfaces, to accommodate the requirements of users with disabilities and to facilitate the deployment of new remote experiments. Another consideration in our work has been to support, as much as possible, the style of learning identiied in Laurillard's Conversational Model 8]. Preliminary feedback indicates that the PEARL architecture facilitates the design and deployment of remote experiments, that such experiments can be embedded in virtual learning environments and that the resulting experiments are well liked by students. This feedback, from both usability and technical viewpoints, has allowed us to improve user interfaces and optimise performance of the system.