This paper presents the first steps of the treatment of a real-world robotic scenario consisting in exploring a large area using several heterogeneous autonomous robots. Addressing this scenario requires combining several components at the planning and execution lev-els. First, the scenario needs to be well modeled in or-der for a planning algorithm to come up with a realistic solution. This implies modeling temporal and spatial synchronization of activities between robots, as well as computing the duration of move activities using a pre-cise terrain model. Then, in order to obtain a robust multi-agent executive layer, we need a robust hierarchi-cal plan scheme that helps identifying appropriate plan repairs when, despite the quality of the various models, failures or delays occur. Finally, we need various algo-rithmic tools in order to obtain flexible plans of good quality.