Exploring Human-Robot Emotional Interaction with Omar Serghini

In this episode, Filippo Sanfilippo hosts Omar Serghini, a PhD researcher specializing in speech emotion recognition, spectrum sensing, and neuromorphic cameras, to discuss the future of human-robot teaming. They delve into how emotional awareness through speech and visual cues can transform collaborative AI systems, especially in healthcare and social contexts.

Key Topics:

• The role of speech-emotional recognition in enhancing human-robot interaction
• How prosody and vocal cues inform a robot’s understanding of human emotions
• Spectrum sensing as a parallel for decision-making under uncertainty in robotics
• The emergence of neuromorphic (event-based) cameras for dynamic scene perception
• Moving from static to dynamic interaction through high-temporal-resolution sensors
• Building multi-channel emotion recognition integrating speech, vision, and context
• Challenges and ethical considerations in developing empathetic machines
• Future prospects of AI in healthcare, elderly care, and emergency rescue scenarios
• The importance of interdisciplinary collaboration for advancing emotional AI

Timestamps:

00:45 - Introduction to Omar Serghini and his research background

01:31 - The importance of emotional detection in human-robot teaming

02:58 - Speech emotion recognition: what it is and why it matters

05:15 - Using vocal cues like tone and rhythm to gauge confidence and hesitation

06:39 - How emotion detection helps robots adapt behavior and ensure safety

08:07 - The evolution of human-robot interaction from command-based to peer-like relationships

10:09 - Spectrum sensing as a metaphor for decision-making in robotics under uncertainty

12:15 - Robustness in noisy environments: lessons from spectrum sensing applied to robotics

16:11 - The role of neuromorphic (event-based) cameras in real-time emotion perception

17:39 - How event-based vision captures fast, subtle changes for dynamic interaction

22:53 - Integrating multiple sensory channels for a richer understanding of human emotions

25:21 - The potential and limitations of robotic empathy in social and care settings

27:22 - Ethical and cultural considerations in emotion recognition technology

28:47 - Challenges and future directions for emotion-aware AI in healthcare and safety

35:23 - Final thoughts on multidisciplinary collaboration and the future of emotional AI

Resources & Links:

• Serghini, O., Serrano, S., Semlali, H., & Maali, A. (2024, September). Robust DNN-Enabled Cooperative Spectrum Sensing. In 2024 International Conference on Software, Telecommunications and Computer Networks (SoftCOM) (pp. 1-6). IEEE.@

Connect with Omar Serghini:

• LinkedIn

Note:

This episode emphasizes the importance of integrating multiple sensory modalities for emotionally intelligent human-robot interactions and highlights ongoing interdisciplinary efforts to bring empathetic AI closer to human behavior.

Insights from Professor Florenc Demrozi. Explore how sensors, AI, and wearable technology are transforming human-robot collaboration, healthcare, and assisted living. Discover the vision of lifelong learning machines and their potential to evolve with us, aiding independence and safety.

Key Topics:

• Evolution of human-robot interaction from isolation to collaboration
• The role of sensors, wearables, and environmental context in human activity recognition
• Challenges in data quality, AI modeling, and system safety for healthcare applications
• The concept of robots as lifelong companions and lifelong learners
• Future scenarios: robots as learning partners, safety enforcers, and co-evolving entities

Timestamps:

00:00 - Introduction to human-robot teaming and interview guest Professor Florenc Demrozi

01:16 - The historical transition from isolated robots to collaborative human-robot interaction

02:13 - The progression towards physical collaboration and shared space in human-robot teaming

02:43 - Recognizing human activity and intent through sensing and AI

03:12 - The importance of understanding human status within physical and environmental contexts

04:08 - How ubiquitous sensors and multi-modal data fusion advance healthcare robotics

05:10 - Applications in healthcare: Parkinson’s, Alzheimer’s, disabilities, and aging population challenges

09:17 - Robots in assisted living: coaching, teaching, and digital therapeutics

11:10 - Empowering the elderly: training and maintaining independence with robots

12:40 - Robots as lifelong companions, physical helpers, and co-learners

14:08 - Challenges of trust, reliability, safety, and real-time collaboration in human-robot teaming

15:02 - Controlling and ensuring safety of robots via constraints and innovative actuators

16:00 - Developing intrinsically safe and elastic actuators like human muscles

17:25 - Data quality, sensor noise, and AI model generalization challenges

20:50 - Regulatory hurdles and the necessity of validation for medical applications

21:51 - Generalization across different environments, sensors, and users

23:21 - Envisioning the future: robots as lifelong learning partners and companions

24:25 - The concept of the lifelong learning digital companion evolving with users

26:25 - Knowledge exchange between humans and machines, and eventual AI surpassing human knowledge in specific tasks

27:24 - Digital coaching systems: recognizing and blocking risky behaviors

28:22 - The vision of humans and robots as equal, continuously evolving partners

29:07 - The role of curiosity and ongoing research in advancing human-robot teaming

29:44 - Closing thoughts and a special message for the next generation of researchers

Resources & Links:

• Florenc Demrozi - University of Stavanger

Connect with Professor Florenc Demrozi:

• LinkedIn