New HRT Taxonomy Framework Organizes 361 Human-Robot Teamwork Requirements

Autonomous systems are increasingly deployed in safety- and mission-critical environments where humans and robots must work together. Yet requirements for human-robot teamwork (HRT) have remained scattered across standards, regulations, and academic papers. To close that gap, a team of researchers led by Anastasia Mavridou and including Hazel Taylor, Sandy Lozito, Louise Dennis, Michael Fisher, and Marie Farrell developed a unified taxonomy of HRT requirements. The team analyzed 361 requirements from 14 cross-domain sources spanning academia and industry, then iteratively classified them into a two-level hierarchy: six high-level categories—information provision, relational control, decision support, safety mechanisms, performance monitoring, and foundational system capabilities—and 21 low-level subcategories. The resulting taxonomy is presented in a paper submitted to the 34th IEEE International Requirements Engineering Conference (RE 2026), available on arXiv (2607.27302).

The taxonomy was validated through two complementary approaches. First, five domain specialists evaluated its structure and completeness. Second, the researchers demonstrated its utility by applying it to an independently assembled corpus of 448 requirements drawn from 19 sources across six HRT domains. This suggests the taxonomy generalizes well beyond its original training data. For engineers and safety analysts, this provides a common vocabulary and checklist for specifying, reviewing, and comparing HRT requirements—critical for ensuring that robotic teammates are predictable and trustworthy. By consolidating fragmented guidance, the framework could streamline certification processes for autonomous systems in healthcare, manufacturing, and defense. It also highlights gaps in current standards, helping standards bodies prioritize future updates. The work marks a step toward mature engineering practices for human-robot collaboration.