TLDR: RobEthiChor is a new approach that allows autonomous robots to make ethical decisions by negotiating based on their users’ personalized ethical preferences and real-time contextual factors. It features a domain-agnostic architecture and a ROS-based implementation (RobEthiChor-Ros). Experiments show it effectively reaches ethical agreements in under a second on average, demonstrating its feasibility and scalability for real-world applications.
As autonomous systems become more integrated into our daily lives, from service robots in hospitals to self-driving cars, a critical challenge emerges: how do these systems make decisions that align with individual human ethical preferences? Unlike humans, who can adapt their moral compass based on context and personal beliefs, current autonomous systems often operate without considering the unique ethical values of their users. This can lead to a lack of trust and a disconnect between the robot’s actions and the user’s expectations.
Introducing RobEthiChor: Ethics-Based Negotiation for Robots
To address this, researchers Mashal Afzal Memon, Gianluca Filippone, Gian Luca Scoccia, Marco Autili, and Paola Inverardi have developed RobEthiChor, an innovative approach that enables autonomous robots to incorporate user ethical preferences and contextual factors into their decision-making through a process of ethics-based negotiation. This system provides a flexible framework for designing robots that can engage in ethical discussions, ensuring their actions are personalized and aligned with the moral beliefs of the people they serve.
RobEthiChor features a versatile, domain-agnostic architecture, meaning it can be applied to various types of autonomous systems. The team also created RobEthiChor-Ros, a practical implementation built on the Robot Operating System (ROS), which can be deployed on actual robots to give them these advanced negotiation capabilities. You can find more details about this research in the full paper available at https://arxiv.org/pdf/2507.22664.
How Does It Work?
The core of RobEthiChor lies in its ability to understand and utilize personalized ethical profiles and real-time contextual information. Here’s a simplified breakdown:
-
User Ethical Profiles: Users can define their ethical preferences, called ‘dispositions,’ and assign grades to them. These grades indicate the importance of a disposition in a given context. For example, a user might prioritize ‘giving precedence to injured people’ differently in an airport versus a hospital setting.
-
Context Awareness: The system continuously monitors environmental factors (like location and time) and the user’s current status (such as being elderly, injured, or in an emergency). This information helps the robot understand the specific situation.
-
Resource Contention: When two or more robots need to use a shared resource simultaneously (like a single elevator or a narrow corridor), a ‘resource contention’ occurs. This is where the negotiation comes into play.
-
Automated Negotiation: Instead of a rigid, pre-programmed decision, the robots engage in a negotiation. They exchange ‘offers’ that include minimal, necessary information about their user’s status. Each robot evaluates the received offer based on its own user’s ethical profile and the disclosed status of the other party. A ‘utility function’ helps quantify the ethical impact of different choices.
A Real-World Scenario: The Airport Dilemma
Imagine two assistive robots, RobAssist A and RobAssist B, at an airport. RobAssist A is helping Alice, an athlete with a leg injury, while RobAssist B is assisting Bob, an elderly traveler who is late for a boarding flight. Both robots need to use the same elevator, which can only accommodate one wheelchair at a time.
Initially, RobAssist B might offer to go first, prioritizing Bob’s elderly status. However, RobAssist A, considering Alice’s injury, might reject this. The negotiation continues, with robots subtly disclosing more information. When RobAssist B reveals Bob’s urgent flight departure and a gate change, RobAssist A re-evaluates. Alice’s ethical profile might include a strong disposition to prioritize others in an emergency, especially the elderly. This benevolent disposition overrides her self-care priority, leading RobAssist A to accept RobAssist B’s offer. Bob and his robot use the elevator first, ensuring he makes his flight, while Alice and her robot wait.
This example highlights how the system enables a ‘contextually ethical’ agreement, where the decision is not fixed but adapts to the specific circumstances and personalized ethical values of the users involved. If no agreement is reached, a default ‘hard ethics’ rule, like first-come-first-served, is applied.
Performance and Scalability
The researchers rigorously tested RobEthiChor-Ros on real robots and in simulated environments. The results were promising:
-
Effectiveness: The system successfully reached ethical agreements in over 73% of scenarios, consistently aligning with the expected outcomes based on user profiles.
-
Efficiency: The negotiation process was remarkably fast, averaging about 0.67 seconds per negotiation, with most completing in under 0.8 seconds. This minimal overhead ensures the system can be practically applied in real-world situations without significant delays.
-
Scalability: The negotiation time increases with the number of negotiation rounds (which depends on the complexity of ethical profiles and disclosed conditions), but it scales approximately linearly. Even in highly complex, stress-tested scenarios (far beyond typical real-world needs), the system demonstrated good scalability, with strategies available to further optimize performance.
Also Read:
- Navigating Global Regulations: A New Logic System for Autonomous Vehicle Design
- Unpacking Societal Values for Ethical AI Decision-Making
Impact and Future Directions
RobEthiChor represents a significant step towards building more trustworthy and socially responsible autonomous systems. By allowing robots to negotiate based on personalized ethical preferences, it fosters greater user trust and enables seamless interaction between systems acting on behalf of different individuals. This approach extends the traditional scope of automated negotiation, moving beyond purely economic or efficiency-driven outcomes to include complex ethical considerations.
Future work aims to expand RobEthiChor to handle multilateral negotiations (involving more than two robots), incorporate time constraints into the negotiation process, and develop more sophisticated models for context and ethical profiles. These advancements will further enhance the ability of autonomous systems to navigate the complexities of human ethics in an increasingly interconnected world.
