Advancing diver-robot interaction capabilities
Project ADRIATIC - Advancing Diver-Robot Interaction Capabilities focuses on enabling cooperative motion between a diver and an autonomous underwater vehicle (AUV). The lecture is going to present novel interaction methods such as controlling the AUV using a gesture recognition glove and modalities of remote testing in challenging COVID times done through the EUMarineRobots TNA framework.
Cooperative geolocalization and CNN-based Automatic Target Recognition for underwater autonomy
Matteo Bresciani and Leonardo Zacchini
The presentation will describe the activities conducted during the LAUA and DAMOSS2 TNA projects. While the LAUA project focused on comparing different techniques for robot localization, with the final goal to apply them in a scenario for the geo-localisation of CO2 bubbles’ sources, DAMOSS2 aimed at investigating innovative automatic strategies for detecting and mapping the seabed CO2 seeps using Autonomous Underwater Vehicles (AUVs). Besides, the DAMOSS2 project sought to evaluate the acoustic localization performance in the presence of CO2 bubbles. These two strictly connected research activities have been conducted using heterogeneous assets to assess the performance of the various localization approaches and different Convolutional Neural Networks (CNNs) to detect and localize the CO2 seeps.