In this thesis we present an experimental research platform in robotics, XPROB. This platform has been designed to be a tool that facilitates the development of robotic applications. XPROB achieves a flexible prototyping system that features a task-level programming environment, a dynamic representation of the work-cell's equipment, and sensor data integration at runtime allowing on-line program monitoring and adaptation.

This thesis describes how the object-orientation paradigm combined with a traditional layered-control structure lead to an open and dynamic architecture. It also presents an advanced object representation to handle high-level reasoning, even about partially recognized objects.

The platform was first evaluated using simple robotic applications, such as assembly and sensor-guided actions. Afterwards, an industrial application, consisting of a disassembly line for worn-out electric motors, was successfully set up and controlled by our platform.