A new generation of UAVs is coming that will help improve the situational awareness and assessment necessary to ensure quality data collection, especially in difficult conditions like natural disasters. Operators should be relieved from time-consuming data collection tasks as much as possible and at the same time, UAVs should assist data collection operations through a more insightful and automated guidance thanks to advanced sensing capabilities. In order to achieve this vision, two challenges must be addressed though. The first one is to achieve a sufficient autonomy, both in terms of navigation and of interpretation of the data sensed. The second one relates to the reliability of the UAV with respect to accidental (safety) or even malicious (security) risks. This however requires the design and development of new embedded architectures for drones to be more autonomous, while mitigating the harm they may potentially cause. We claim that the increased complexity and flexibility of such platforms requires resorting to modelling, simulation, or formal verification techniques in order to validate such critical aspects of the platform. This paper first discusses the potential and challenges faced by autonomous UAVs for data acquisition. The design of a flexible and adaptable embedded UAV architecture is then addressed. Finally, the need for validating the properties of the platform is discussed. Our approach is sketched and illustrated with the example of a lightweight drone performing 3D reconstructions out of the combination of 2D image acquisition and a specific motion control.