One of the key aspects of the Internet of Things revolution is the integration of digital components within the physical world. Examples range from autonomous vehicles (e.g. good delivering robots) to large scale sensor-actuator networks, to monitor and possibly control environmental and artificial systems (e.g. smart buildings, seismic and water quality monitoring). Such systems are also known as Cyber-physical systems (CPSs). Our focus will be on the design of large-scale distributed CPS acting in a spatial environment subject to unpredictable changes. The main challenge is to handle the underlying uncertainty during the design phase. The aim is to develop techniques for assessing behavioural guarantees for CPSs, mitigating the risk of design flaws. The final goal is to create a model predictive control infrastructure, allowing us to include spatio temporal properties, stochastic models and the environment, and to leverage information from the provided measurments of uncertainty.