Next-Gen Tactile Sensing for Companion and Care Robots to Debut at ICRA 2026

Text
Researchers at the University of Ulsan (UNIST) presented a breakthrough in companion and care robotics at the 2026 International Conference on Robotics and Automation (ICRA 2026), according to a report by Edaily. The technology enables robots to interpret tactile interactions—such as petting or hand-holding—as emotional signals, marking a significant advancement in human-robot interaction.

The development, led by a team at UNIST’s Robotics and AI Laboratory, uses advanced pressure-sensing arrays and machine learning algorithms to differentiate between casual contact and emotionally charged touch. When a human strokes a robot’s head or grasps its hand, the system analyzes pressure patterns, duration, and movement to infer emotional intent. This capability, described as “a critical step toward more intuitive robotics,” could enhance applications in healthcare, education, and elderly care.

“Current robots lack the ability to ‘feel’ or interpret human touch beyond basic physical contact,” said Dr. Ji-Hoon Kim, a UNIST researcher involved in the project. “Our system translates tactile feedback into emotional context, allowing robots to respond with empathy rather than programmed routines.” The technology was demonstrated using a prototype humanoid robot, which adjusted its behavior—such as tilting its head or emitting a calming hum—based on the perceived emotional tone of human interactions.

The research was unveiled during ICRA 2026, one of the world’s leading robotics conferences, which drew over 5,000 attendees from academia, industry, and government agencies. While UNIST did not disclose commercialization timelines, the team emphasized that the technology could be integrated into existing care robots within the next five years.

Edaily reported that the project received funding from the South Korean Ministry of Science and ICT, which has prioritized AI-driven robotics as part of its national innovation strategy. The ministry’s chief innovation officer, Dr. Hwang Sun-kyu, stated the technology aligns with efforts to address labor shortages in eldercare by deploying emotionally responsive machines.

Critics, however, raised questions about the ethical implications of robots mimicking human empathy. “While the technical achievement is impressive, we must ask: What happens when a machine’s ‘empathy’ becomes indistinguishable from human care?” asked Dr. Min Seo-jin, a bioethicist at Seoul National University. The UNIST team acknowledged these concerns, noting that their system is designed to complement—not replace—human interaction.

The project builds on earlier work by UNIST researchers, who previously developed tactile sensors for robotic surgery and prosthetics. This latest iteration represents a shift toward social robotics, a field that has seen rapid growth since the 2020s. According to a 2025 report by the International Federation of Robotics, the global market for care robots is projected to exceed $12 billion by 2030, driven by aging populations and increasing demand for personalized assistance.

While the Edaily article did not provide details on the technology’s testing phases, a separate statement from ICRA 2026 organizers highlighted the potential for such innovations to redefine human-robot relationships. “This is not about replacing humans but expanding the possibilities of how we support each other,” said Dr. Laura Chen, a robotics engineer and conference co-chair.

The UNIST team plans to publish a peer-reviewed study on the technology in the Journal of Artificial Intelligence Research later this year. Until then, the project remains a focal point of discussion at ICRA 2026, where attendees debated the future of emotionally intelligent machines.

Text
Subheading
How Does the Technology Work?

The system relies on a combination of flexible tactile sensors and neural networks trained on thousands of human touch interactions. These sensors, embedded in the robot’s surface, detect pressure, temperature, and shear forces. Machine learning models then process this data to classify touch as friendly, anxious, comforting, or aggressive. For example, a slow, deliberate stroke might be interpreted as soothing, while a quick, jerky motion could signal distress.

Text
Subheading
What Are the Implications?

The technology’s applications extend beyond care robots. UNIST researchers are exploring its use in educational settings, where robots could adapt to a child’s emotional state during learning sessions. Additionally, the system could improve disaster response robots, allowing them to gauge human distress through physical contact. However, the team cautioned that widespread adoption would require rigorous safety and ethical reviews.

Text
Subheading
What’s Next?

UNIST has partnered with South Korean tech firms to test the technology in real-world scenarios, including nursing homes and outpatient clinics. The project’s success could influence global robotics standards, particularly in countries with aging populations. Meanwhile, the broader robotics community remains divided on the balance between technological advancement and human-centric design.

Text
Quoted textSource
“Emotional touch detection is a game-changer for robotics,” said Dr. Ji-Hoon Kim, UNIST researcher. “It bridges the gap between mechanical functions and human-like understanding.”