Predicting Environmental Boundary Behaviors With a Mobile Robot

Predicting the behavior of dangerous environmental boundaries, like spreading fire or oil spill, provides relevant information to mitigate the problem or even to support evacuation actions in order to save human or animal lives. In this letter, we present a model that uses a single robot moving around an environmental boundary in order to predict its shape by an analytical continuous function, which is based on the combination of polynomial approximation and Fourier Series. We show that the method converges to the exact boundary when we increase the sample frequency and the robot velocity. In order to evaluate the estimation quality, we performed experiments with simulated and actual robots. We applied our model in some dynamic boundaries presented in the literature, as in the application of plume-front estimation, showing that it accomplish accurate results.