PILLAR-Robots and Human-Robot Interaction

The PILLAR project aims to create a new generation of robots with a higher level of autonomy, capable of determining their own goals and strategies. However, what impact could the interaction between humans and robots (HRI) have on the experience that robots acquire and develop over their lifetime to satisfy the needs of their human designers and users in real-world applications? Let’s examine some interesting parameters we encounter in the PILLAR project.

  • Purpose: the heart of PILLAR-Robots

Robot autonomy is crucial for ensuring versatility and adaptability, but Open-Ended Learning (OEL) robots frequently encounter a significant challenge: their tendency to become distracted by any interesting experience they come across. To address this, our robots’ hearts beat with the rhythm of purpose. Purpose represents what the designer or user wants from the robot, ensuring that the robot’s learning and actions are focused and relevant. By encoding these purposes into an internal motivational framework called desire, robots can direct their exploration toward acquiring knowledge that aligns with these goals. This approach makes desires domain-independent, allowing robots to adapt to various environments by linking their desires to specific, domain-dependent goals [1].

As a result, OEL robots’ purpose-driven approach affects HRI, and vice versa. Specifically, having a not arbitrary learning process but guided by relevant goals and objectives means that interactions between humans and robots become more meaningful and efficient, as robots are better equipped to understand and align with human intentions and goals. Additionally, robots can more effectively integrate feedback from human interactions, which is crucial for refining and improving their actions, leading to faster and more effective learning cycles. A purpose-driven approach also allows robots to maintain their operational focus despite these changes, improving their ability to adapt to and respond to new challenges autonomously. This dynamic adaptation capability enhances HRI by ensuring that robots remain reliable and effective partners in diverse scenarios.

Let’s take a closer look at the PILLAR project’s three use cases. What are the advantages of HRI in developing PILLAR-Robots?

Agri-food Application: Humans may provide real-time input to robots on sorting quality and size criteria, allowing robots to adapt to new varieties of product and changing quality standards. Furthermore, as human operators add additional fruits and vegetables or change sorting criteria, robots will use this feedback to gradually learn and enhance their sorting algorithms, resulting in greater performance over time.

Edutainment Application: Since this application allows for social engagement, HRI’s impact on robotic advancement is limitless. For starters, robots can interact with students and educators to collect data on individual learning styles, preferences, and progress, allowing for more personalized educational experiences. Robots can keep students engaged and motivated by observing and responding to their emotional and social cues. Teachers and students can provide feedback on the robot’s instructional content and procedures, enabling robots to improve their teaching strategies and efficacy.

Industry/Retail Application: Robots can collaborate with human workers, learning from their approaches and obtaining direction on complex tasks to improve their operational procedures. Humans can also provide feedback on task priorities and changes in the workplace, allowing robots to adjust their activities in real time to suit changing production demands.[1] Baldassarre, Gianluca, Richard J. Duro, Emilio Cartoni, Mehdi Khamassi, Alejandro Romero, and Vieri Giuliano Santucci. “Purpose for Open-Ended Learning Robots: A Computational Taxonomy, Definition, and Operationalisation.” arXiv preprint arXiv:2403.02514 (2024).

Leave a Reply

Your email address will not be published. Required fields are marked *