In this project, we present SPECTRA - Strategic Protocol Evaluation and Configuration Testbed for Responsible Autonomy which is designed as a framework to compare autonomous agents with different models of ethics. It is built on a network representing a multi-agent setup, where different ethical agents interact with each other over time. The framework presents a responsibility dilemma i.e. a scenario with conflicting choices, one which is individually beneficial for the agent and the other which is individually suboptimal for the agent but good for the system or the collective.
Using the SPECTRA framework, we model ethical agents driven by different paradigms of ethics like Deontology, Virtue Ethics, Utilitarianism and a recently introduced paradigm of Computational Transcendence.