Probabilistic modelling and verification using RoboChart and PRISM
Kangfeng Ye, Ana Cavalcanti, Simon Foster, Alvaro Miyazawa, Jim Woodcock
Abstract
Abstract RoboChart is a timed domain-specific language for robotics, distinctive in its support for automated verification by model checking and theorem proving. Since uncertainty is an essential part of robotic systems, we present here an extension to RoboChart to model uncertainty using probabilism. The extension enriches RoboChart state machines with probability through a new construct: probabilistic junctions as the source of transitions with a probability value. RoboChart has an accompanying tool, called RoboTool, for modelling and verification of functional and real-time behaviour. We present here also an automatic technique, implemented in RoboTool, to transform a RoboChart model into a PRISM model for verification. We have extended the property language of RoboTool so that probabilistic properties expressed in temporal logic can be written using controlled natural language.