Edificios inteligentes y computación afectiva para mejorar la interacción humano robot

Estado: 
Número de proyecto: 
834-C2-145
Vigencia:
De 07/Mar/2022 hasta 06/Mar/2025

Objetivo:

Construir una solución tecnológica que permita que un edificio inteligente facilite la interacción afectiva entre humanos y robots


Descripción:

El objetivo de este proyecto es integrar la capacidad de un edificio inteligente de producir información y la posibilidad de uno o más robots de aprovechar esta información para mejorar su interacción con los humanos. A modo de ejemplo, el caso más sencillo de interacción al que aspira este proyecto es que un robot pueda ir a un laboratorio a saludar a personas que acaban de ingresar. Esto solo es posible si el edificio inteligente es capaz de notificar al robot que hay personas en un espacio determinado. El edificio también podría detectar la presencia de ruido en el ambiente (por ejemplo de un grupo de personas) y determinar si hay otras personas en la cercanía (como en una aula) y consultarles si les molesta el ruido, si desean cerrar la puerta, o si desean que el robot mismo converse con las personas ruidosas para mejorar el ambiente. Luego, es posible aspirar también a casos de uso más complejos, como por ejemplo que el edificio pueda además identificar a la persona y conocer sus preferencias (si toma café o té, o si suele entrar al edificio a horas precisas). Un robot también podría estar en la entrada de un edificio y guiar a las personas que ingresan hacia un laboratorio o auditorio donde se lleve a cabo una actividad, tras inferir en tiempo real sobre dicha actividad basado en la información que provean ciertos sensores en el edificio. Por último, una tercera aspiración del proyecto, más ambiciosa y novedosa, es que el edificio sea capaz de identificar ciertas emociones de sus habitantes y que el robot modifique su comportamiento en función de eso.

Investigador principal
Dr. Adrian Lara Petitdemange

Colaboradores
Dra. Kryscia Ramírez Benavides
Dr. Luis Quesada Quirós
Dr. Allan Berrocal Rojas
Dr. Adrian Lara Petitdemange
Bach. Diego Orozco Fonseca

Unidad académica base
Centro de Investigaciones en Tecnologías de la Información y Comunicación (CITIC)

Unidades académicas colaboradoras
Escuela de Ciencias de la Computación e Informática (ECCI)

Publicaciones asociadas

Understanding Students' Perspectives About Human-Building Interactions in the Context of Smart Buildings

Descripción:

Smart buildings provide a variety of sensor-based services to support and enhance the quality of human activities. Advanced technologies such as robotics are increasingly added to smart buildings’ ecosystems, creating a need to incorporate affective computing techniques to augment the quality of human-building, and human-robot interactions. To better understand user’s needs and expectations about human-building interactions, we conducted a pilot study using a mixed methods approach combining short surveys and controlled laboratory activities. We recruited 66 participants and collected several data elements characterizing their perceptions and expectations about smart building services. This paper presents preliminary evidence showing acceptance of specific human-building interaction methods based on ambient-sensors information such as in-context voice, behavior, and emotion, recognition. We also identified a need for educational activities to promote the understanding of smart building concepts and their impact in modern society. These results can be leveraged to assist the design of future services that include human-building and human-robot interactions.

Tipo de publicación: Book Chapter

Publicado en: Proceedings of the 15th International Conference on Ubiquitous Computing & Ambient Intelligence (UCAmI 2023)

Transfer Learning and Fine-Tuning for Facial Expression Recognition with Class Balancing

Descripción:

Facial expression recognition benefits from deep learning models because of their ability to automatically extract features. However, these models face three important challenges: first, training tends to take longer times than with traditional machine learning models. Second, obtaining and labeling enough data samples can become a heavy burden due to the feature complexity usually involved in these problems. Third, it is also common to face class imbalance challenges. In this paper, we address these challenges by implementing transfer learning, oversampling and fine tuning to a facial expression recognition use case. Combining transfer learning with the use of a GPU helped us complete the training for our models in just about one hour. Furthermore, we achieved a 65.75% accuracy with one of the models. We provide measurements for metrics that are helpful when dealing with imbalanced data to assess that the models are not biased like precision, recall, F1 score and loss.

Tipo de publicación: Conference Paper

Publicado en: 2024 L Latin American Computer Conference (CLEI)

Real-time malicious URL detection

Descripción:

Malicious URLs are constantly used for phishing, malware distribution and other illegal activities. Because benign URLs are needed for the Internet to function, malicious URLs are hard to block. While several works have focused on offline classification of malicious URLs, real-time detection still needs to be investigated. This paper evaluates the performance of real-time malicious URL detection using two techniques: blacklist methods and machine learning methods, deployed in both local and cloud environments. The study highlights significant differences in latency and connection failure rates under various load conditions, providing insights into the strengths and limitations of each approach. The blacklist method consistently demonstrates lower latency, making it suitable for scenarios requiring quick response times, though its stability may be compromised under high loads in a local setup. In contrast, the machine learning method offers advanced detection capabilities but exhibits higher latency, particularly in local environments, due to its resource-intensive nature. The cloud environment mitigates some latency issues but still lags behind the blacklist method in terms of speed. The findings emphasize that most latency stems from the verification process, with the local environment requiring significant optimization to reduce delays. The study concludes that implementing a proxy for real-time URL detection is viable, especially in cloud environments, where resource management can better handle increased demand.

Tipo de publicación: Conference Paper

Publicado en: 2024 IEEE Latin-American Conference on Communications (LATINCOM)

Taxonomy of Malicious URL Detection Techniques

Descripción:

Malicious URLs are often used by phishing campaigns, botnets and other attacks. Indeed, DNS traffic is necessary for the Internet to function correctly, which means that this data flow cannot be blocked. For these reasons, detecting malicious URLs is both important, challenging and still an open research problem. There are two types of techniques used to detect malicious URLs: rules-based and machine learning-based. The traditional, rules-based techniques rely on blacklists and heuristics. These techniques struggle to keep up with a rapidly changing array of malicious URLs. Therefore, machine learning-based techniques have emerged. Both detection techniques rely on URL characteristics such as length, number of vowels and others to classify them as legitimate or malicious. The main contribution of this paper is to propose a taxonomy of detection techniques and to point out which URL characteristics are used by each method. While surveys on the topic exist, a precise mapping between the detection methods and the characteristics is not available. We also compare these techniques, highlighting that machine learning-based techniques are more complex to implement but better at keeping up with rapidly incoming new malicious URLs. In contrast, rules-based techniques are simpler and easier to implement, but they struggle to update fast enough to identify new malicious URLs.

Tipo de publicación: Book Chapter

Publicado en: Information Technology and Systems