Probabilistic Model-Based Programming Techniques for Prediction, Analysis and Control (PROMPT)

Model-based programming (MBP) languages such as SysML and RMPL provide declarative descriptions of the structure, processes, functions, and context of a system. They are particularly useful for cyber-physical systems that interact with the environment through sensors and actuators. MBP languages are beneficial both in the design of systems and in the control of systems in real time. However, existing MBP languages are deterministic and do not model uncertainties that influence the systems behavior. We propose to develop Probabilistic Model-Based Programming Techniques for Prediction, Analysis and Control (PROMPT). PROMPT will provide probabilistic extensions to SysML, enabling that to enable the representation of uncertainties in both the structure and the behaviors of a complex system, as well as interactions between the structure and behaviors. PROMPT will provide inference services for predicting the behavior of a system under uncertainty and estimating the current state of the system from noisy sensors. These inference services will be used to predict performance and analyze models at design time and to control a cyber-physical system at runtime based on probabilistic beliefs about the state of the system. PROMPT will be applied to and evaluated on a significant cyber-physical system such as an unmanned undersea vehicle.