NSF (GAČR) STRoLL

The project aims to develop methods for spatio-temporal environment mapping and vision-based life-long localization and navigation. The primary objective is to enable long-term autonomous operation of mobile robots in naturally changing outdoor environments. Long-term autonomy will be achieved through combination of methods that explicitly model the environment dynamics with robust self-localization and navigation techniques. The spatio-temporal representations will model uncertainties of the environment states by probabilistic functions of time. This will allow to continuously integrate new observations into the environment model, learn about the dynamics of environment changes and use the obtained knowledge to predict the environment states for a given time. This predictive capability will improve robustness of self-localization and efficiency of planning that will allow to obtain more informative environment observations and improve the spatio-temporal model.