Ethical and safe learning for deploying multi-robot systems in complex real-world environments
SECOND CALL
APPLICATION ID: ALL17
What we are looking for:
We seek innovative research proposals that explore how to design learning environments for robots so that the policies they learn to accomplish their tasks are guaranteed to be safe and ethical, even when multiple robots are involved. This is a paradigm shift that focuses on the learning environment, instead of on the learning algorithm employed by each robot, to guarantee the properties that robots’ behaviours must exhibit.
The context:
AI research is being challenged with ensuring that autonomous agents, such as robots, learn behaviours that are both ethical (in alignment with moral values) and safe.
The problem to address:
Ángela Ribeiro (CAR), Maite López-Sánchez and Juan A. Rodríguez-Aguilar (IIIA) join efforts to address the problem of learning safe and ethical behaviour policies for robots in charge of agricultural tasks.
Objectives:
- To investigate the main ethical and safety constraints involved in agricultural robotics that can be operationalised in a prototipical scenario.
- To develop safe and ethical learning environments for a single robot for different agricultural tasks that involve humans.
Qualifications:
-
Indispensable: PhD on AI or Robotics
-
Strong background in deep learning
-
Strong background in mathematics
-
Programming skills in Python and C
Expected Outcomes:
- A taxonomy of ethical values and safety constraints in the domain of agricultural robotics.
- Innovative AI techniques to automate the design of safe and ethical environments for learning robots in agricultural robotics.
