Edge-AI Powered Real-Time Emotion Recognition in Social Robots for Healthcare Companionship

Abstract

The recognition of emotions has become a fundamental part of human-robot interaction in contemporary society, notably in healthcare contexts where social robots may be used as companions, assistants and therapeutic companions. The conventional cloud-based emotion recognition solutions are also limited by latency, privacy issues, and dependency on reliable internet connectivity, which are not acceptable in hospitals, elderly care facilities, and home-based health care settings where responsiveness and data security are the main priorities. The Edge-AI technology eliminates such drawbacks by allowing emotion detection models to execute directly on-device, which has a great impact on response time and is certain to make sure that emotional data that is sensitive does not go out of the immediate environment of the user. The study examines the abilities to include lightweight neural networks, multimodal sensing, and real-time inference pipelines into healthcare companionship social robots. The research highlights the model efficiency, emotion profiling of users as individual, and dynamic learning, which is needed, in long-term interaction with patients whose emotional state changes periodically with age, illness, isolation, etc.Over the past several years, healthcare robotics have become focused on empathetic interaction where the system must be able to detect minor emotional expressions including micro-expressions, voice tremors, and behavioral changes. The recognition of emotions using edge-AI systems increases this ability by subjecting intricate visual and audio cues to streamlined deep learning designs, such as MobileNet, TinyML frameworks and quantized convolutional networks. This solution will enable the robots to react instantly to any discomfort, depression, anxiety, or loneliness in a patient, and thus enhance psychological well-being and clinical monitoring. The current study assesses the performance of the system in various healthcare conditions, which include design issues, including the robustness of the models, constant learning, and sensor noise, as well as the suggestion of the methods of including emotion-sensitive robots into the process of patient care. Results indicate that Edge-AI-enabled robots do not only enhance emotional engagement, but also decrease the workload and enable caregivers to foster holistic patient care with trustworthy and real-time emotionally intelligent and privacy-sensitive products.