DART project has been structured in four Work Packages following a layer-based approach. The
core WP1 will provide especially designed datasets to the remaining WPs. Upon this WP, the WP2
basically evaluates the suitability of proposed machine learning techniques to elucidate which is
the best alternative to enable robust and accurate data-driven trajectory prediction capabilities, and
under which conditions. This WP2 will make use of the outputs of WP1 and provide inputs to WP3.
The WP3 will leverage the outputs from previous WPs to devise and evaluate a mechanism for
detecting the influence of surrounding traffic on a trajectory prediction and enhancing the prediction
capabilities of algorithms towards a collaborative trajectory prediction process. This mechanism,
based on collaborative reinforcement learning techniques will account for network complexity
effects and will return updated predictions by considering the complexity of the actual ATM
environment. Finally, WP4 will provide the project management activities required for the overall
coordination of the defined WPs, including the dissemination and project impact activities.
