A large body of work is coming together to show how autonomous systems might be ethically supervised by humans.
Abstract. Recent experience has shown that the concept of robot ethics is important for establishing norms defining allowed behaviors for unmanned systems. However, approaches considered to date are either based on highly abstract artificial intelligence schemes or else uniquely “hard wired” into a given robotic architecture in an unrepeatable fashion. A more-general approach is needed for defining and deploying ethical constraints on robotic systems that are expected to operate autonomously with only occasional human control.
This paper explores development of a practical approach to ethical operation of unmanned maritime systems in maritime environments. This approach is based on the Rational Behavior Model (RBM), a three-layer robot control architecture modeled on the control hierarchy of Naval vessels. RBM utilizes a finite state machine as the basis for mission definition and high-level control. This approach provides for exhaustive pre-mission testing by human operators and predictable runtime decision-making that takes mission-specific ethical constraints into account.
The authors believe that these achievable approaches for ethically constrained mission planning and operations can be applied to a broad range of unmanned systems. The potential legal and ethical impacts of unmanned systems in the real world must be considered and addressed. Although our focus is on the most challenging circumstances of autonomous lethality in naval scenarios, the same approaches also appear to be relevant to scientific, commercial and civilian operations by unmanned maritime systems. This work opens a practical door leading into that necessary future.
Recent and related work: