TDM: Trustworthy Decision-Making Via Interpretability Enhancement

Human-robot interactive decision-making is increasingly becoming ubiquitous, and trust an influential factor in determining the reliance on autonomy. However, it not reasonable to systems that are beyond our comprehension, typical machine learning data-driven black-box paradigms impede interpretability. Therefore, critical establish computational trustworthy mechanisms enhanced by interpretability-aware strategies. To this end, we propose a Trustworthy Decision-Making (TDM) framework, which integrates symbolic planning into sequential decision-making. The framework learns interpretable subtasks result complex, higher-level composite task can be formally evaluated using proposed metric. TDM enables subtask-level interpretability design converges optimal plan from learned subtasks. Moreover, TDM-based algorithm introduced demonstrate unification of with other sequential-decision making algorithms, reaping benefits both. Experimental results validate effectiveness trust-score-based while improving