Software de respuesta empática para robots

Abstract

This undergraduate project aims to develop software that enables a social robot (such as the Pepper robot available at Santo Tomás University) to respond empathetically to different human emotions, focusing primarily on joy and sadness. The idea stems from a clear need: current robots still fail to establish a genuine emotional connection with people, which greatly limits their role insocial environments. To achieve a more natural and engaging interaction, the software was designed based on a conceptual model of empathy, allowing the robot to identify emotions through three main channels: facial recognition, voice recognition, and keyword detection. Each channel was assigned a specific weight within a weighted voting system, prioritizing the most reliable one according to psychological studies and internal testing. Development was carried out using ROS, connecting an external node to Pepper to ensure effective communication without fully relying on the robot's system. During controlled testing, the system successfully recognized emotions with good accuracy, particularly in detecting joy. However, some challenges were also observed, such as sensitivity to lighting, noise, or accessories like glasses, which can affect facial recognition. Despite these limitations, the system was proven functional and met the goal of delivering a more empathetic interaction. This project not only strengthens the research line of the GED Group in social robotics but also opens the possibility of applying such solutions in emotional support or mental health (under the supervision and guidance of qualified professionals).