19 April 2022 in IEEE Transactions on Neural Networks and Learning Systems
We have developed a smart dive glove that recognizes 13 static hand gestures used in diving communication. The smart glove employs five dielectric elastomer sensors to capture finger motion and implements a machine learning classifier in the onboard electronics to recognize gestures. Five basic classification algorithms are trained and assessed: the decision tree, support vector machine (SVM), logistic regression, Gaussian naïve Bayes, and multilayer perceptron. These basic classifiers were selected as they perform well in multiclass classification problems, can be trained using supervised learning, and are model-based algorithms that can be implemented on a microprocessor. The training dataset was collected from 24 participants providing for a range of different hand sizes. After training, the algorithms were evaluated in a dry environment using data collected from ten new participants to test how well they cope with new information. Furthermore, an underwater experiment was conducted to assess any impact of the underwater environment on each algorithm's classification. The results show all classifiers performed well in a dry environment. The accuracies and F1-scores range between 0.95 and 0.98, where the logistic regressor and SVM have the highest scores for both the accuracy and F1-score (0.98). The underwater results showed that all algorithms work underwater; however, the performance drops when divers must focus on buoyancy control, breathing, and diver trim.
6 April 2024 in Computer Networks
14 June 2023 in IEEE Journal of Oceanic Engineering
Despite the popularity of the buddy system, where a pair of divers look out for each other, scuba diving still poses significant risks with 15–30 deaths per 100 000 divers per year. In the 2015 edition of “ Diving Medicine for Scuba Divers ,” it is reported that 86% of diving fatalities involve a lone diver, yet divers still elect to dive alone for a variety of reasons including buddy unavailability. A solution to this may lie in a dedicated autonomous dive-buddy robot, which could also operate as an assistant. The challenge is how would the diver interact and communicate with the robot in underwater conditions often involving low visibility. In this article, we detail and demonstrate a technological solution that involves the direct motion capture of fingers and hand using a smart dive glove with integrated stretch sensors alongside an inertial measurement unit. The hardware and gesture recognition protocol were initially evaluated in a lab-based study where 10 users participated in a virtual reality environment setup and interacted with a virtual robot using gestures. In total, 140 out of 175 gestures were correctly recognized, resulting in diver-robot commands. A full sea demonstration of the underwater motion-capture system followed with the underwater dive glove used to send gesture-based acoustic commands to a dedicated dive buddy, a modified autonomous underwater vehicle. The diver was able to command the robot to move in six directions with a total delay of 6 s or less between the beginning of the gesture and the initiation of the action by the robot.
14 October 2022 in Marine Technology Society Journal
Diver‐robot interaction is an exciting and recent field of study. There are different ways a diver and robot can interact, such as using tablets or detecting divers with cameras or sonars. A novel approach presented in this paper uses direct diver‐robot communication. To facilitate communication for humans, we use typical diver gestures, which are transmitted to a robot using a wearable glove and acoustic communications. Following previous work by the University of Zagreb and the University of Auckland, a collaboration to control an autonomous underwater vehicle based on a wearable diver glove has been made possible through the EU Marine Robots project. Under this project, Trans-National Access trials allow Laboratory for Underwater Systems and Technologies, University of Zagreb, to offer its robots and infrastructure to external partners. Initial trials with the University of Auckland, which were planned to take place on site, were transformed into remote access trials. This paper reports on these challenging trials and collaboration given the distance and time zone difference. The key point is to demonstrate the possibility of having a diver remotely controlling a robot using typical gestures recognized by a wearable glove and transmitted via acoustic modems (and the Internet for the remote connection).