In May 2014, off the southern coast of Portugal, an international multi-vehicle control experiment involving aerial, surface and underwater vehicles was attempted.

Read more at http://sunfish.lsts.pt/

Involved researchers:

Javier Gilabert, UPCT, Spain
Tor Arne Johansen, NTNU, Norway
Kanna Rajan, FEUP, Portugal
Joao Sousa, FEUP, Portugal

PERSISTS' (Persistent Autonomous Aerial, Surface and Underwater Vehicles tracking Sun Fish) key objectives were threefold: to demonstrate the state of the art (and practice) in coordinated measurements using diverse assets looking at the same patch of the ocean at the same time. Two, to provide high resolution in-situ data for scientific observations to foster inter-disciplinary science and engineering. Three, to provide predictive (modeling) capability to understand the biological dynamics of Mola mola (bony Sunfish). And finally to generate interest in inter-disciplinary science in the next generation of researchers in STEM (Science, Technology, Engineering and Mathematics).

The principal science objective was to work with biologists to tag and track Mola's and provide contextual environment data. While Mola's have been tracked by this science team over the last four years, obtaining high-resolution in-situ science data with the tags in use, has been challenging. A critical need was to understand why the Mola's follow a specific trajectory in the water-column (a ‘Mola corridor') and what kinds of environmental conditions the Mola are, in turn, tracking. Our engineering objectives dove-tailed with the need to demonstrate the capability to bring together diverse autonomous assets with advanced methods in Artificial Intelligence (AI) based deliberative control for such observations, using a suite of sophisticated ground-based command and control tools being used at FEUP/UPorto.

The intent was to conduct a two week experiment off the coast of Olhao in the Algarve. This region was chosen because of the presence of a commercial Tuna pen, where the Mola's are often trapped providing biologists with the opportunity to tag and release these fish in open waters. New inexpensive tags, designed at FEUP, were built and derived from of two different technologies. Our experiment design called for a UAV to be launched from a research vessel, once an indication of a Mola on the surface was received. The UAV would be sent to the vicinity of the tag location, visual identification was to follow over an IP-video feed, leading to the deployment of an AUV with CTD sensors as well as the retargeting of a WaveGlider autonomous surface platform. Both the UAV and the AUV are controlled by the Teleo-Reactive EXecutive (T-REX) an AI-based controller onboard the AUV and on the ground for UAV control.

Weather conditions coupled with issues related to ship capabilities did not allow for launch and recovery operations as planned; however separate deployments of UAVs from shore and AUV and WaveGlider deployments from ship to survey the ‘Mola corridor' to make coordinated measurements between an AUV and the WaveGlider, detection of a thermal gradient in the water column near a shelf leading to frontal zone entrapping biota such as salp and the testing of the integrated tool chain distributed between onboard and off board robotic components in open waters were the key outcomes. Finally, the outreach effort with 8th graders in a school in the Porto region resulted in substantial interest in kids intrigued by the science and technology behind the experiment, which in turn lead to a Mola capture and dissection in class, a first for the school.