Verifiable autonomy and responsible robotics
|Dates:||6 March 2019|
|Times:||14:00 - 15:00|
|What is it:||Seminar|
|Organiser:||Department of Computer Science|
|Who is it for:||External researchers, Adults, Alumni, Current University students, General public|
Join us for this School research seminar, part of the Intelligence and autonomy seminar series and hosted by Professor Angelo Cangelosi.
Robots will increasingly make their own decisions about what to do and when to do it without necessarily consulting humans. How can we trust these systems if we do not know what they will choose to do and, crucially, why they make these choices?
In this talk, guest speaker Professor Michael Fisher will describe how the use of rational agents at the core of autonomous systems forms the basis for not only transparency and explainability, but verifiability of behaviour and decision-making. Once we have systems based on such hybrid agent architectures, then we can strongly analyse them and move towards not only trustworthiness, but safe, ethical and responsible behaviour.
He will provide some examples from his team's current and previous projects examining the deployment of hybrid agent architectures in both robots (domestic and industrial) and autonomous vehicles (road and air), covering the verification of safety, ethics, fault tolerance, etc.
This work, and the examples described, derive from projects funded by EPSRC: https://autonomy-and-verification-uol.github.io/projects
Role: Professor of Computer Science
Organisation: University of Liverpool
Travel and Contact Information