The general goal is to develop planning, coordination and communication tools for the distributed control of a team of aerial and aquatic robots for time-extended monitoring of marine environments. Ultimately, the aim is to exploit autonomous robot teams to allow frequent and pervasive data gathering in Gulf waters. We will consider a 3D team featuring autonomous robots with complementary sensory-motor capabilities: Unmanned Aerial Vehicles (UAVs) (multi-rotors) and Unmanned Surface Vehicles (USVs). Robots are medium-sized, equipped with sensors for environmental monitoring, and GPS devices. The team features multiple vehicles, aiming to create a model of robot swarm, to cover large areas by parallel deployment of multiple, spatially distributed resources. The system will be designed for use in long-running missions, that would allow repeated, time-indexed data sampling. In order to deliver full autonomy, long-running missions, and effective area coverage the research will develop distributed and scalable solutions for controlling individual and system-level behaviors. The output of the project will consist of: 1) Algorithms for online action planning, to optimize the outcome of the monitoring operations. A robot plan specifies a list of tasks, including: portions of the environment where to perform monitoring actions, what type of actions, for how long. Plans are autonomously computed and adapted by each robot based on mission goals, encountered issues, or reacting to new evidence (e.g., newly sampled data indicate that a certain area should be sampled more than previously planned). Solution approaches will rely on integer programming models for automated planning, coupled with auction-based methods. Robots make use of a wireless communication network to exchange information about plans, fuse sample data, and negotiate on task assignments. 2) Algorithms for coordinated control and plan execution. A plan gives a coarse-grained description of actions. The precise implementation of the plan, including motion actuation, will be dealt by control-based approaches, whose task is to define trajectories that: are collision-free, maximize the gathered mutual information, support fairness of coverage, intensify sampling actions on the most interesting portions of the area. Data exchanged through the network plays a main role for the effective distributed coordination of the controls. 3) Network and mobility control for data transmissions in the robot MANET. Since a data communication infrastructure cannot be guaranteed in general mission scenarios, communications have to rely on a Mobile Ad hoc NETtwork (MANET) established among robots. Information exchange will rely on periodic robot broadcast of profile and mission information. Network connectivity will be supported by: inclusion of robot proximity constraints in planning and coordination models; specification of move actions to maintain/improve/repair connectivity; control on transmission power to adapt wireless ranges. Threshold-based approaches from swarm intelligence will be used to schedule network decisions. 4) Automatic takeoff/landing for UAV/USV + Rendez-vous scheduling. Since commercially available UAVs usually have much reduced power autonomy with respect to USVs, the idea here is to let UAVs to be carried by USVs to recharge or save energy. This will make long-term missions possible, overcoming current battery limitations for drones. Robust algorithms for the online scheduling of UAV/USV rendez-vous for landing operations will be developed. At the same time, robust control procedures will be devised to manage landing and takeoff operations. The task is made difficult by mutual relative motion in 6 DOFs, restricted area of landing platform, and the need for precise mutual localization. Note that the recharge station for the UAV will not be implemented in the real robots, being a task requiring custom hardware development. The developed techniques for planning, control, coordination, and networking will have general applicability and will be supported by theoretical results as well as empirical evaluation, both in simulation and in realistic testbeds. Two demonstrators, one Qatar and one at the ISME test facilities in Italy, will show integrated final results. The demonstrators will be based on a proof-of-concept team of up to 6 USVs and two UAVs at work in a controlled aquatic environment. All the approaches and solutions 1-4 outlined above will contribute to state-of-the-art with novel algorithms and methods for planning and coordinated control problems in distributed robotics. Control of takeoff/landing operations under unstable conditions is a difficult task to achieve and no standard solutions exist. Similar considerations apply to robust scheduling under uncertainty. The decisional components described in 1-4 (planning, coordination, motion control, networking scheduling) will be treated as fully interdependent, such that a core innovation will be the design of a distributed architecture that makes all these components work in closed-loop interaction with each other. The research team includes PIs with similar yet complementary characteristics that fully cover the different expertise needed for the research. All PIs have participated or are participating to large European projects strongly related to the proposal. The Italian co-PIs are part of ISME, a university consortium for research on marine technologies with a focus on robotics. The participation of ISME will position the project in an international context and will allow to share some of ISME’s testing and robotic resources. The project will output algorithms / control technologies with a TLR around 5, and will pave the road to further developments with more powerful robotic technology, that could be ready for deployment in Gulf waters. It would allow to move from sporadic (and costly) sampling missions to repeated, extensive, and goal-directed data sampling.