Deciphering Robo-Emotions: Unraveling the Path to Artificial Empathy
SECTION 1: SOCIAL ROBOTS
Introduction
In a groundbreaking leap from science fiction to reality, the world of emotional recognition using social robots is redefining human-robot interactions. From joy to sadness, anger, and surprise, emotions are at the heart of the human experience. This blog delves into the transformative realm of emotion recognition, where social robots not only perform tasks but also engage with us in ways that mirror human responses, fostering natural and harmonious interactions.
Overview
This article will cover:
1. Paradigm of Emotional Recognition in Social Robots
2. Emotional Robot Models
3. Invoking Emotions in Robots
4. Methodology (with case-study)
5. What’s Next?
6. Challenges and Ethics
7. Conclusion
8. References
1. Paradigm of Emotional Recognition in Social Robots
Emotion recognition in the context of social robots is the capability to perceive and interpret human emotions through sensors and algorithms, enabling robots to respond empathetically and create more human-like interactions. Emotions have been considered from three main points of view as follows:
I. Formalisation of the robot’s own emotional state: Modelling emotion states in robots by defining neuro-computational models.
II. Emotional expression of robots: Studies focused on exploring which modalities (e.g., face expression, body posture, movement, voice) can convey emotional information from robots to humans and how people perceive and recognise emotional states.
III. Ability of robots to infer the human emotional state: Recent works aim to design algorithms for classifying emotional states from different input modalities, such as facial expression, body language, etc. are enabling robots to understand human emotions.
2. Emotional Robot Models
Traditionally, there are three types of emotional robot models:
• Neurobiological Models: Emotions are modelled through computational constructs based on neural structures.
• Psychological Models: Emotions are conveyed through expressive actions and behaviours.
• Data-Driven Models: These are task-specific robots with intent planning, behaviour planning, and realisation.
SECTION 2: EMOTIONAL INTELLIGENCE IN SOCIAL ROBOTS
3. Invoking Emotions in Robots
Robots can create the illusion of life through their physical presence and movements. Our perception of a robot’s emotions is often based on how it moves; for example, backing away from a door might be interpreted as avoidance and fear. In more complex interactions, like in emotional robots (ERs), rich, goal-oriented behaviour is essential to convey believable responses. Explicitly modelling emotions can influence not only actions but also cognitive processes such as reasoning, planning, and learning.
• Selecting the appropriate architecture is key to evoking emotions in robots, as it integrates drives and emotions to guide their behaviours and decision-making.
• Choosing the right theoretical framework provides the foundation for triggering emotional states in robots within a computational system.
• Enhancing robot aesthetics includes mechanisms like muscle mimicry, colour choices, and expressive features, all designed to improve emotional expressiveness based on functionality.
4. Methodology
Emotion recognition in social robots involves three main technical aspects:
4.1 Data Collection:
• Data in sensory (touch sensors), audio (speech and tone of voice) , visual (facial expressions, body language, and gestures), textual form is collected according to the functionality of the model.
• It is then cleaned, eliminating garbage information and values.
4.2 Data Processing and Feature Extraction:
• A framework for evaluation of the model is chosen. A popular one being,
SAIBA (Situation, Agent, Intention, Behaviour, Animation). Features best fitting to generate the output required are extracted.
• Data processing can me done using softwares such as Microsoft Azure (facial recognition), Praat (speech manipulation), Gazebo (testing robot’s sensory capabilities in stimulated circumstances)
• Taking the example of the SAIBA Framework, it is divided into 3 parts:
I. Intent Planning: It is to determine the user’s intent. Intent planning involves analysing the user’s actions, words, and other cues to understand what they want or need from the robot.
II. Behaviour Planning: Here, the ER decides how it should respond to best meet the user’s needs and expectations. It considers the user’s emotions and intent in determining the most suitable course of action.
III. Behaviour Realisation: Now the robot executes the behaviour it has planned, whether it’s offering comforting words, providing information, or engaging in any other action designed to cater to the user’s emotional state and intent.
4.3 Sentiment Analysis and Machine Learning Algorithms:
• The machine learning algorithms then analyse the data from these sensors to identify patterns that are associated with different emotions. For example, the algorithm might learn that a smile is associated with happiness, a frown is associated with sadness, and a raised voice is associated with anger.
• Once the robot has recognised the user’s emotion, it can use this information to adapt its behaviour accordingly.
Case-Study Example
AIM: Sophia, the first anthropomorphic robot to express over 60 human feelings/emotions, was selected to examine the interplay between the emotional expressions produced by robots and consumers’ affective responses using machine learning methods. This study measured robotic emotional expressions in terms of anger, contempt, disgust, fear, happiness, neutral, sadness, and surprise, whereas the consumers’ affective reaction was measured by positive, negative, and neutral sentiments.
SECTION 3: REALITY CHECK
5. What’s Next?
We are entering an amazing new era for robotics, AI and Affective Computing because concrete emotional robots are being developed to a level at which they can be deployed in social settings. Advancing this robot-human interaction will enhance communication personalise interactions, which in turn opens multiple doors, enabling robots to help out in therapy and collaborative professional work, extending emotional depth of virtual interactions, security and law enforcement, market research and advertising and business negotiations.
6. Challenges and Ethics
The advent of emotionally intelligent robots, like any new technological innovation, doesn’t come without its concerns and rightfully so.
The main ethical concerns relating to emotional robots are:
• Privacy and Data Security
• Emotional Manipulation
• Dependency and Addiction
• Aggravating loss of employment
• Indeterminate moral code of conduct
Building trust strategies:
• Promoting transparency in the capabilities and limitations of ERs and providing insights into the decision-making processes of ML algorithms.
• Granting users control over their interactions with ERs through customisable behaviours, boundaries, and easy opt-out options.
• Incorporating ethical design principles focusing on user well-being, privacy, fairness, transparency, and accountability throughout the development process.
• Encouraging continuous user feedback and integrating it into the development process.
• Establishing regulatory frameworks and oversight mechanisms to govern the ethical use.
Conclusion
In culmination, emotional recognition with social robots is in its early stages, with the potential for revolutionising human-robot interactions. Despite ethical concerns and challenges, the right approach can unlock a transformative future in how we interact with technology.
References
https://www.frontiersin.org/articles/10.3389/frobt.2020.532279/full#T1
https://www.tandfonline.com/doi/abs/10.1080/0952813X.2023.2263456?src=
https://www.sciencedirect.com/science/article/pii/S096969892100117X
https://askwhy.medium.com/robots-as-emotional-sentient-being-challenges-ahead-5d2db54956a0
