Unlocking the Secrets of Qatar’s Deep Waters
A novel AI-based approach for creating information maps for marine environments
Qatar is home to a very diverse marine environment and its preservation carries a direct impact on the economic, social, and environmental activities in the country. Therefore, a robust system is crucial to ensure reliable and efficient mechanisms to collect, analyze, and process data collected from marine and aquatic environments.
Information maps that can provide useful information about the marine ecosystem, including maps of water depths, water quality, and distribution of flora and fauna, are an integral part of such a system. In marine environments, the common practice currently for such map building is to send out a single, manually operated, relatively large boat equipped with powerful sensors on missions that can last for days, sampling data at locations specified beforehand.
This is a very costly process that poses several challenges, and due to varying factors, cannot be always relied on to produce accurate and dependable results. To solve these challenges, a team of researchers led by Dr. Gianni A. Di Caro, an associate teaching professor at Carnegie Mellon University in Qatar, has successfully developed TARMEM – a solution aimed at defining innovative and efficient solutions for building information maps by monitoring and surveying marine and aquatic environments.
Funded under Qatar National Research Fund’s National Priorities Research Program (NPRP-10-0213-170458), TARMEM’s solution concept radically departs from the traditional approach by using Artificial Intelligence (AI) solutions and features the use of teams of heterogeneous autonomous robots. These teams comprise of Unmanned Surface Vehicles (USVs), which are relatively small-sized robot boats, and Unmanned Aerial Vehicles (UAVs), which are multi-copter flying robots. To ensure that UAVs can execute long-duration map building missions, the UAVs are carried on top of the USVs for battery recharging.
Instead of a single, powerful, and centralized system, TARMEM uses a distributed fleet, or swarm, of relatively low-cost, unmanned robots with different sensory-motor capabilities, and work out their coordination and cooperation to achieve effective system-level synergies. Using the swarm approach boosts efficiency and improves accuracy in map building and information-updating activities, as robots can coordinate to keep sampling data in parallel, at different locations, using sensors with complementary competencies. Moreover, a swarm provides inherent redundancy against failure or issues of a few units.
However, obtaining these advantages is not that simple and comes at the cost of the challenges posed by a highly complex mechanism for controlling and coordinating the actions of the system. Keeping this in mind, the research team has successfully developed a number of novel approaches that provide innovative solutions for distributed data sharing and fusion, online computation, and execution of team-level plans, control of collision-free navigation, provisioning of network connectivity, and seamless UAVs/USVs interaction.