Optimal Cooperative Control of Mobile Sensors for Dynamic Process Estimation

A typical mission for robotic systems in environmental monitoring is the identification of dynamic processes like atmospheric dispersion by a group of mobile sensors. Due to this problem’s large-scale and highly dynamic character, an efficient cooperative sampling strategy is required. This paper presents a mathematical concept for estimating the parameters of a Gaussian puff model of the dispersion process based on cooperatively gathered measurement data. The sensors’ cooperative behavior is determined by a distributed model-predictive controller. It combines task allocation and trajectory planning in a single mixed-integer problem formulation employing linearly approximated vehicle dynamics models. By integrating the quality of the dispersion model parameters as an additional optimization criterion for the controller, the proposed method intends to simultaneously solve both the estimation problem and the cooperative control problem in a near optimal manner.