Long-term Trustworthy Autonomous Systems

The Universities of Oxford, Leeds, and York, with University College London, are delighted to announce that their collaborative project RAILS (Responsible AI for Long-term Trustworthy Autonomous Systems) is one of three projects funded as part of the Trustworthy Autonomous Systems (TAS) Hub’s Responsibility call. Led by Dr Lars Kunze, Departmental Lecturer at the Oxford Robotics Institute (Department of Engineering Science) and Technical Lead at the  Responsible Technology Institute, the project focuses on the challenges associated with the long-term operation of autonomous systems and the effects of change on these systems. In particular, it will examine two main challenges that have been little-studied: (i) open-ended dynamic environments and (ii) systems that change over time.

The project aims to:

  • Draw up a responsibility ‘map’ after evaluating public and expert views of autonomous systems.
  • Create a cloud-based model system for evaluating the causal responsibility and accountability of autonomous systems that can persist after the end of the project and be used within the TAS network and RAS community
  • Develop an assurance case for the safe operation of lifelong learning and self-adaptive autonomous systems in post-deployment settings
  • Design an adaptive governance framework for the continuous development and long-term deployment of autonomous systems

Dr Kunze commented, “The increasing prevalence of autonomous systems demands that we look ahead to the challenges that might stem from changes in those systems once they deploy, as well as investigate the impact of highly likely changes in their environment. With this project we hope to both draw attention to these challenges and suggest some possible ways forward to address them.”

The project includes Professor Marina Jirotka at the University of Oxford, Professor Jack Stilgoe at University College London, Dr Jo-Ann Pattinson at the University of Leeds, and Dr Richard Hawkins at the University of York. The project is proud to be working with a range of partners across industry and civil society – collaborators include Imperium Drive Ltd, Addleshaw Goddard LLP, Oxbotica, TLR Ltd, the Autonomous Drivers Alliance (ADA), the United Nations’ ITU-T Focus Group on AI for Autonomous and Assisted Driving, the Society for Computers & Law, and AWS. It will also collaborate with other projects within the TAS Hub.