From 31 August to 11 September 2015, the Center of IMAR of the University of the Azores hosted the final field trials of a new type of robotics-based distributed sensor system, aimed at affording marine scientists and commercial operators a revolutionary tool for marine habitat mapping in complex underwater 3D environments. The trials form the ultimate project demonstration of the EC research project Marine Robotic System of Self-Organising, Logically Linked Physical Nodes (MORPH).
What is so special about the MORPH Supra Vehicle is that its building blocks (a set of small robotic marine vehicles) operate in a cooperative manner, as if they were a single, powerful vehicle, capable of carrying advanced, complementary sensor suites, and reacting to environmental conditions in real time.
Launched in February 2012, the 4-year research project MORPH is funded by the European Union's Seventh Framework Program and Industry with a budget of 8,5 million Euros. The major goal of the project is to develop a distributed robotic underwater system carrying complementary sensors for complex underwater mapping missions. At the core of the innovative solution adopted is the concerted operation of a number of relatively simple, affordable vehicles, capable of cooperating seamlessly and behave as an ensemble, yielding a “MORPH Supra Vehicle”.
Research engineers, marine biologists, and commercial partners have been working together to meet the formidable challenge of providing scientific and commercial end-users an advanced mapping system capable of accessing inhospitable areas and operating under scenarios that defy current technology; e.g., to perform underwater surveys over high-relief seafloor with complex 3D structures (canyons and rugged cliff areas) in the presence of reduced visibility and natural unforeseen obstacles.
To prove the concept of a MORPH Supra Vehicle, the tests took place in the beautiful setting of “Baia de Porto Pim”, in Faial Island (Azores). The MORPH ensemble, or Supra Vehicle, is a fleet of 3 to 6 distinct vehicles, that operate together to perform a video and sonar data survey to map an underwater cliff face and adjacent sea-bottom, dynamically adapting to the geometry of the survey area.
The research activities of Jacobs University within the MORPH project are led by Prof. Andreas Birk. He and his team have developed algorithms for automatic 3D perception and detailed 3D mapping in complex surroundings or so-called “unstructured environments”. Underwater environments pose a particular challenge for the generation of detailed maps due to the fact that there is no GPS available and it is very hard to localize the robot. Therefore, a technique known as registration is the main method to determine the spatial relations between the sensor data: the appearance of the sensor data is used to stitch sequences of the data together, e.g., natural landmarks are detected in images and used as reference points to combine all images in a map. The team from the Jacobs University has developed new methods for the registration of 2D and 3D data as well as for the optimization of maps with a process known as Simultaneous Localization and Mapping (SLAM). These new methods are very fast and very robust, and hence well suited for the challenging conditions of underwater operations. The work of Jacobs Robotics also featured aspects of semantic mapping that were included in the Azores trials, namely in form of a classification of the terrain complexity, which was used to adapt the speed of the vehicles to cover more interesting terrain more densely, and in form of a machine classification of terrain deformations by stingrays to autonomously estimate the size of the stingray population in the surveyed area.