Second Roboethics Competition: What can a robot fetch for you?
ICRA 2022 just got started and this year the second Roboethics Competition will take place as part of this conference! This year, the competition received five submissions representing participants from Canada, United Kingdom, and the United States, demonstrating an increase over the single submission from last year’s competition at ROMAN 2021. We are very excited to see the growth in interest and engagement for roboethics from students and researchers!
This year’s competition has two parts. In the first part, the participants were asked to submit an ethics proposal where they outline a home robot’s ethical decision making policy with regards to what items it can fetch and deliver. The robot’s main task is to fetch objects and deliver it to a specific person or location. In the second part, participants (new ones are welcomed!) take part in a hackathon where they can choose to implement one of the submitted ethics proposals.
Competitions can transform roboethics discussions into practical design problems.”
(Rhim et al. 2022)
Tell us what you think!
We want to know what you think about this year’s submissions and what you would consider if you were to buy a home robot someday.
Please fill out this short survey to share your thoughts!
Competition submissions
The following is a list of the five submissions from this year and a brief description of each:
TEAM 1: Value Trained Fetch-Bot: “The Fetch-Bot is a deep reinforcement learning robot that is equipped with a reward function that optimizes the robot to complete fetch tasks for items located in all permissible areas of the house and return them to the requestor.”
TEAM 2: Risk Ratings Database for Ethical Decision Making (R2D2): “R2D2 is an ethical solutions framework based on a data driven situational risk assessment module to provide quantifiable metrics to support the ethical decision-making processes.”
TEAM 3: Emotion and Context Aware Behaviour Trees: “Robot’s ethical policy is implemented through a behavioural tree, where the robot is required to check if it has any knowledge of the request such as previous requested item, identity of the human, emotion of the human and context information of the environment.”
TEAM 4: EthioFetch: “EthioFetch checks if it is safe for the requested object to be delivered to the destination (location/ person) and if the destination and requester both have the permission to access the requested object.”
TEAM 5: Ethical Home Robot: “The Ethical Home Robot’s policy follows three sections: 1) check the access rules of the object and the cognitive state of the requestor, 2) follow safety protocols during task execution, 3) listen for feedback afterwards.”
