Robotic System Specification Methodology Based on Hierarchical Petri Nets

The presentation concerns a model-based approach to developing robotic system controllers. Central to this approach is a parameterised  meta-model that describes the generic robotic system from two points of view: structure and activity. By appropriate parametrisation of the meta-model one can obtain a particular model of a robotic system performing desired tasks. The meta-model is expressed using the Robotic System Hierarchical Petri Net (RSHPN), a 6-layer Petri net tailored for robotics. Each layer describes the activity of the robotic system at a completely different level of abstraction. This guarantees the separation of concerns. The required model emerges from the meta-model by appropriate parametrisation of the layers of the RSHPN. Introduction of parametrisation enables the robot designer to focus only on the concepts derived from the domain. It greatly facilitates the robotic system development process as it gives the designer clear guidance on what needs to be defined and what is imposed by the design pattern. The resulting single RSHPN model is used both to verify some properties of the system, e.g. lack of deadlocks, but also to automatically generate controller code. The presented approach is illustrated by examples of the creation of two different robotic systems.