Policy sandboxes for autonomous transportation in controlled environments: Evidence and insights from open-pit mining operations

While the deployment of autonomous vehicles in complex urban settings remains constrained by safety bottlenecks and regulatory conservatism, controlled environments that are characterized by spatial confinement, centralized management, and high operational demand, have emerged as ideal sandboxes for accelerating policy and technological maturity. Among these, Autonomous Haulage Systems (AHS) in open-pit mining is one of the most mature and real-world implementations, recently demonstrated by the milestone delivery of over 300 Komatsu-brand AHS trucks to Rio Tinto. This maturity offers critical, yet underutilized, insights for bridging the gap between technological capabilities and governance frameworks. To this end, this paper aims to trace the evolution of knowledge, technology, and governance, and the future development in this field to inform broader autonomous transportation policies, through bibliometric analysis and topic modeling over 171 core research works that are published since 2015. The findings show that AHS development is embedded within a deeply interconnected academia–industry ecosystem, where innovations advance through iterative co-regulation rather than isolated experimentation typical of urban autonomous driving. Research history analysis reveals a definitive shift from a hardware-centric focus to cyber-physical system integration, specifically prioritizing interoperability standards, human–automation collaboration and application scalability. Moreover, the governance of these systems follows a "stepwise autonomy" pathway, where operational boundaries are expanded incrementally based on demonstrated performance, underpinned by risk-based oversight, structured safety validation, and communication robustness. The derived policy insights can inform other controlled environments (e.g., ports, airports, and industrial parks), while requiring targeted adaptation to mixed-traffic interfaces, safety exposure, temporal constraints, and institutional fragmentation. Overall, the study highlights how controlled environments can function as policy sandboxes that accelerate both technology maturity and regulatory learning toward safe and scalable autonomous mobility.