Tutorial 5 intro: CNR-INM – Citizen engagement: learning by imitation in marine robotics
“SWAMP learning from humans” experiment @ BTS 2021. The active involvement of citizens and stakeholders from the early stages is fundamental in research projects on the development of autonomous physical agents able to operate in presence of and to cooperate with human beings and manned systems as, for instance, in the case of Unmanned Marine Vehicles working in harbour and coastal areas. Thanks to its innovative mechanical design (pump-jet propulsion, foam-made hull, etc.), SWAMP, a modular reconfigurable lightweight Autonomous Surface Vehicle for ultra-shallow water applications, constitutes an easily deployable and safe platform for carrying out experiments involving citizens and students. A first example was given by the “SWAMP learning from humans” experiment, where citizens attending Festival della Comunicazione in Camogli, Italy, remotely piloted SWAMP to execute a simple task consisting in crossing a couple of doors delimited by buoys in 2019 and 2021. Commands and trajectories recorded while the vehicle executed the goal task remotely controlled by citizens trained an AI control system based on neural networks. Thanks to the control law generated by this learning by imitation system, the robot was then able to autonomously navigate executing the goal task as learnt from people. The tutorial will consist of a general introduction to the topic of learning by imitation in marine robotics followed a replica of the experiment at sea involving BTS participants piloting SWAMP ASV. Suggestions for further experiments and citizens/students engagement in CNR-INM “learning by human” marine robotics research will be encouraged.