Researchers from the Social Robotics Group at the Robotics Lab of Universidad Carlos III de Madrid have created an AI-driven system that enables a companion robot to identify the people it interacts with. The prototype, called Mía, is currently undergoing trials in Madrid City Council day centers to provide personalized emotional support for older adults experiencing cognitive decline.
One of the key hurdles in social robotics is enabling robots to engage with people in a natural, individualized way. A common solution is to equip robots with the ability to identify who they are interacting with. To address this, researchers at the Social Robotics Group of the Robotics Lab at Universidad Carlos III de Madrid developed a voice-recognition system that runs directly on Mía, a rabbit-shaped companion robot.
According to researcher José Carlos Castillo Montoya, the technology is designed to support emotional stimulation for older adults with cognitive decline, with user identification serving as a key step toward tailoring the robot’s responses to each person’s needs.
Unlike conventional recognition systems that depend on cameras and image analysis, the new approach relies solely on the robot’s built-in microphone. The system processes data locally on the device, enhancing privacy by keeping voiceprints on the robot itself while also allowing it to learn and recognize users progressively through ongoing interactions.
“The robot doesn’t require a separate training phase to learn a user’s voice. It can create a voice profile automatically the first time someone speaks to it,” said Arecia Segura Bencomo from the Department of Systems Engineering and Automation at Universidad Carlos III de Madrid.
Advantages Of Animal-Like Robotics and Emotional Stimulation
This development, integrated into Mía—a rabbit-shaped robotic prototype fully created at Universidad Carlos III de Madrid—belongs to the field of animal robotics. This therapeutic approach aims to replicate the well-documented benefits of animal-assisted therapy, such as lowering stress and anxiety, while working within controlled settings where live animals may not be suitable due to welfare concerns or limitations in patients’ ability to provide care.
According to José Carlos Castillo Montoya, in cases involving older adults with cognitive decline, the robot encourages users to engage in caregiving behaviors. Pilot studies conducted in Madrid City Council day centers show that beyond improving mood, the robot also serves as a social facilitator—helping reduce isolation, easing feelings of loneliness, and prompting greater interaction among users.
An AI-Powered “Voice Fingerprint”
The system works in a way that mirrors human learning. Each time a person speaks to the robot, the algorithm examines subtle acoustic features in the voice to build a distinctive “voice signature.”
José Carlos Castillo Montoya explained that the approach is intentionally lightweight due to the robot’s limited processing power. “The algorithm creates a kind of map and, using improved open-source code developed in our lab, it clusters similar signatures together. Each cluster ends up representing a specific user.”
The system is also designed to handle new situations in real time. When it encounters an unknown speaker, it recognizes that the voice does not match any existing group. If interaction continues, it can form a new profile dynamically, effectively learning a new user. In tests, the software demonstrated strong accuracy in identifying familiar individuals.
This development paves the way for a new generation of more accessible assistive and companion robots that can enter a home and automatically learn who lives there simply by hearing a first greeting like “good morning.” The research team emphasized the long-term potential of this level of personalization.
José Carlos Castillo Montoya noted that the goal is for the robot to adjust its behavior depending on who it is interacting with. “If it recognizes a particular user who tends to be more restless, it can automatically activate specific mechanisms and behaviors designed to help calm that person.”
At present, the team is testing multiple working prototypes of Mía in real clinical environments to further refine the technology, while also exploring partnerships with companies interested in transferring the system into commercial assistive robotics platforms.
Read the original article on: Tech Xplore
Read more: This autonomous robotic toilet drives itself to your location