Verifiable Autonomy and Responsible Robotics

Abstract

The move towards greater autonomy presents challenges for software engineering. As we may be delegating greater responsibility to software systems and as these autonomous systems can make their own decisions and take their own actions, a step change in the way the systems are developed and verified is needed. This step involves moving from just considering what the system does, but also why it chooses to do it (since decision-making may be delegated). In this chapter, we provide an overview of our programme of work in this area: utilising hybrid agent architectures, exposing and verifying the reasons for decisions, and applying this to assessing a range of properties of autonomous systems.

Verifiable autonomy: From theory to applications

“…While at Liverpool the group worked with artists Cecile B. Evans to produce the "Sprung a Leak" exhibit at Tate Liverpool during 2016 and 2017 7 . This is an 18 minute play, involving three robots, as well as synchronised video images; members of the public could walk within the play as it unfolded.…”

“…Our development of programming languages has proceeded from executable (temporal) logic specifications [2,3], through resource-boundedness and deliberation [4,5] and on to to Beliefs-Desires-Intentions (BDI) languages explicitly developed with verification [6,7] or ethical reasoning in mind [8,9].…”