Generative Adversarial Networks and Data Clustering for Likable Drone Design.

Novel applications for human-drone interaction demand new design approaches, such as social drones that need to be perceived as likable by users. However, given the complexity of the likability perception process, gathering such design information from the interaction context is intricate. This work leverages deep learning-based techniques to generate novel likable drone images. We collected a drone image database (N=360) applicable for design research and assessed the drone's likability ratings in a user study (N=379). We employed two clustering methodologies: 1. likability-based, which resulted in non-likable, neutral, and likable drone clusters; and 2. feature-based (VGG, PCA), which resulted in drone clusters characterized by visual similarity; both clustered using the K-means algorithm. A characterization process identified three drone features: colorfulness, animal-like representation, and emotional expressions through facial features, which affect drone likability, going beyond prior research. We used the likable drone cluster (N=122) for generating new images using StyleGAN2-ADA and addressed the dataset size limitation using specific configurations and transfer learning. Our results were mitigated due to the dataset size; thus, we illustrate the feasibility of our approach by generating new images using the original database. Our findings demonstrate the effectiveness of Generative Adversarial Networks (GANs) exploitation for drone design, and to the best of our knowledge, this work is the first to suggest GANs for such application.

Generating Alerts from Breathing Pattern Outliers.

Analysing human physiological data allows access to the health state and the state of mind of the subject individual. Whenever a person is sick, having a panic attack, happy or scared, physiological signals will be different. In terms of physiological signals, we focus, in this manuscript, on monitoring breathing patterns. The scope can be extended to also address heart rate and other variables. We describe an analysis of breathing rate patterns during activities including resting, walking, running and watching a movie. We model normal breathing behaviours by statistically analysing signals, processed to represent quantities of interest. We consider moving maximum/minimum, the amplitude and the Fourier transform of the respiration signal, working with different window sizes. We then learn a statistical model for the basal behaviour, per individual, and detect outliers. When outliers are detected, a system that incorporates our approach would send a visible signal through a smart garment or through other means. We describe alert generation performance in two datasets-one literature dataset and one collected as a field study for this work. In particular, when learning personal rest distributions for the breathing signals of 14 subjects, we see alerts generated more often when the same individual is running than when they are tested in rest conditions.

First Step toward Gestural Recognition in Harsh Environments.

We are witnessing a rise in the use of ground and aerial robots in first response missions. These robots provide novel opportunities to support first responders and lower the risk to people's lives. As these robots become increasingly autonomous, researchers are seeking ways to enable natural communication strategies between robots and first responders, such as using gestural interaction. First response work often takes place in harsh environments, which hold unique challenges for gesture sensing and recognition, including in low-visibility environments, making the gestural interaction non-trivial. As such, an adequate choice of sensors and algorithms needs to be made to support gestural recognition in harsh environments. In this work, we compare the performances of three common types of remote sensors, namely RGB, depth, and thermal cameras, using various algorithms, in simulated harsh environments. Our results show 90 to 96% recognition accuracy (respectively with or without smoke) with the use of protective equipment. This work provides future researchers with clear data points to support them in their choice of sensors and algorithms for gestural interaction with robots in harsh environments.

Integrating drones in response to public health emergencies: A combined framework to explore technology acceptance.

The aim of the study was to propose and test an integrated model combining the technology acceptance model (TAM), task-technology fit (TTF), social motivation, and drone-related perceived risks to explore the intention to use drones in public health emergencies (PHEs). We conducted a survey among the Israeli population, yielding a sample of 568 participants. Structural equation modeling was implemented to test the research hypotheses. The results showed that our integrated model provided a robust and comprehensive framework to perform an in-depth investigation of the factors and mechanisms affecting drone acceptance in PHEs. First, ease of use, attitudes, individual-technology fit, task-technology fit, and social influence significantly and directly influenced users' behavioral intention to utilize drone technology. Second, attitudes were significant mediators of the effects of social influence and perceived risks on the intention to use drones. Finally, significant relationships between TAM, TTF, social motivation, and perceived risks were also observed. Theoretical aspects and practical implications-which can serve as the basis for shaping a positive development in drone public acceptance in PHEs and in general-are discussed.