Fleet-Level Autonomous Drone Logistics Dataset

The dataset captures fleet-level operational telemetry of autonomous drone logistics collected from a large-scale port-centric logistics environment operating under continuous, real-world conditions. The data span a period from March 3, 2022, to August 3, 2025, recorded at a 10-minute temporal resolution, reflecting sustained day-to-day logistics activity across multiple operational seasons, weather regimes, and workload intensities. The dataset is structured as a multivariate time-series, where each record represents the instantaneous state of an active drone within a coordinated fleet, along with contextual information describing tasks, infrastructure load, environmental conditions, and inter-drone interactions. Operational measurements include drone kinematics and energy states, task and delivery context, fleet coordination indicators, port infrastructure utilization, environmental and temporal conditions, and safety-related risk signals. These features are derived from onboard flight controllers, fleet coordination middleware, port infrastructure monitoring systems, and environmental sensing units routinely deployed in modern intelligent logistics hubs. The dataset reflects naturally imbalanced operational patterns, such as peak-hour congestion, rare safety events, uneven zone utilization, and long-tailed energy consumption behaviors, which are characteristic of real logistics environments. In addition to raw operational features, the dataset provides fleet-level learning targets that quantify coordination effectiveness and system stress. These include the Fleet Coordination Efficiency Index (FCEI), Airspace Stress Index (ASI), and Coordination-Induced Energy Savings (CIES), enabling direct investigation of how coordinated decision-making influences efficiency, safety, and sustainability at scale. An estimated time-of-arrival variable is also included as an auxiliary benchmark signal. The dataset is released in CSV format with clearly defined feature semantics and temporal ordering, making it suitable for regression-based learning, temporal modeling, and coordination analysis in intelligent logistics research. It is publicly shared via Zenodo to support reproducibility, benchmarking, and future extensions in autonomous logistics and industrial cyber–physical systems research