Modeling and tracking of spatio-temporal scalar fields with multiple robots

This paper focuses on evaluation of the sampling strategy under different conditions. We first implement a framework for modeling and tracking spatio-temporal scalar fields with multiple robots, using Gaussian Process, Voronoi partition, discrete and continuous optimization algorithms. Then, the sampling conditions are divided into four cases from two perspectives (asynchronous / synchronous sampling in continuous region / discrete set of locations) and are compared by simulations. The results show that there is no significant difference in the performance of sampling in continuous region and discrete set, but asynchronous sampling could give better prediction accuracy compared to synchronous sampling.