* QNRF-funded project takes novel approach to map-building of marine environments
Carnegie Mellon University in Qatar’s (CMU-Q) Gianni Di Caro is using artificial intelligence (AI) and a fleet of autonomous marine robots to better explore the marine environment around Qatar.
Di Caro, who is leading a team of researchers, is an associate teaching professor at CMU-Q, a Qatar Foundation partner university.
To better understand marine environments, researchers create information maps of data like depth, water quality and salinity. This information is critical for a country like Qatar, which balances offshore oil and gas operations with the preservation and sustainability of a fragile marine ecosystem, CMU-Q has said in a statement.
Typically, information maps are created by sending, every few months, a single, big, manually operated boat to sample data at pre-defined points, one at a time. This method has serious drawbacks, including the fact that sampling is sequential and static. In fact, it doesn’t adaptively select where to sample based on gathered evidence, since data processing is done offline, the statement notes.
Di Caro’s team has rethought the process: “We are using AI to co-ordinate a fleet of unmanned and relatively small aerial and marine robots to gather data over a large area. Using this swarm approach allows sampling data from different locations at the same time and continually and automatically adapting the mission to freshly gathered data, resulting in more accurate maps.”
The project is called Teams of Aquatic/Aerial Robots for Marine Environment Monitoring, or Tarmem, and funded by a grant to CMU-Q from the Qatar National Research Fund (QNRF).
Led by Di Caro, the team also includes Italian investigators Filippo Arrichiello from the University of Cassino and Southern Lazio, and Enrico Simetti from the University of Genova.
While the swarm approach improves efficiency and accuracy in map building, it is a very complex task to co-ordinate a fleet of autonomous air and sea robots. “One immediate complication is that the aerial vehicles run out of power relatively quickly,” said Di Caro. “We have, therefore, designed the system so the robot boats also serve as carriers and recharging stations. This enables us to execute missions over longer periods of time, in fully unattended modality.”
Artificial intelligence methods are used throughout the system to co-ordinate the complexity of multiple autonomous vehicles collecting and sharing data while also navigating unpredictable environmental conditions. AI methods allow for the online distributed planning of sampling activities, collision-free navigation while maintaining network connectivity, distributed data sharing and seamless interaction between the robots in the air and those in the water.
Because of this extensive, innovative use of AI methods, Tarmem was selected for the QNRF Research Outcome Seminar on Artificial Intelligence, and the QNRF 'Research Matters' newsletter.
The multi-year project will next focus on validation and outdoor testing in marine environments.
Di Caro, who is leading a team of researchers, is an associate teaching professor at CMU-Q, a Qatar Foundation partner university.
To better understand marine environments, researchers create information maps of data like depth, water quality and salinity. This information is critical for a country like Qatar, which balances offshore oil and gas operations with the preservation and sustainability of a fragile marine ecosystem, CMU-Q has said in a statement.
Typically, information maps are created by sending, every few months, a single, big, manually operated boat to sample data at pre-defined points, one at a time. This method has serious drawbacks, including the fact that sampling is sequential and static. In fact, it doesn’t adaptively select where to sample based on gathered evidence, since data processing is done offline, the statement notes.
Di Caro’s team has rethought the process: “We are using AI to co-ordinate a fleet of unmanned and relatively small aerial and marine robots to gather data over a large area. Using this swarm approach allows sampling data from different locations at the same time and continually and automatically adapting the mission to freshly gathered data, resulting in more accurate maps.”
The project is called Teams of Aquatic/Aerial Robots for Marine Environment Monitoring, or Tarmem, and funded by a grant to CMU-Q from the Qatar National Research Fund (QNRF).
Led by Di Caro, the team also includes Italian investigators Filippo Arrichiello from the University of Cassino and Southern Lazio, and Enrico Simetti from the University of Genova.
While the swarm approach improves efficiency and accuracy in map building, it is a very complex task to co-ordinate a fleet of autonomous air and sea robots. “One immediate complication is that the aerial vehicles run out of power relatively quickly,” said Di Caro. “We have, therefore, designed the system so the robot boats also serve as carriers and recharging stations. This enables us to execute missions over longer periods of time, in fully unattended modality.”
Artificial intelligence methods are used throughout the system to co-ordinate the complexity of multiple autonomous vehicles collecting and sharing data while also navigating unpredictable environmental conditions. AI methods allow for the online distributed planning of sampling activities, collision-free navigation while maintaining network connectivity, distributed data sharing and seamless interaction between the robots in the air and those in the water.
Because of this extensive, innovative use of AI methods, Tarmem was selected for the QNRF Research Outcome Seminar on Artificial Intelligence, and the QNRF 'Research Matters' newsletter.
The multi-year project will next focus on validation and outdoor testing in marine environments.