TY - EJOU AU - Li, Xiao-Juan AU - Zhang, Yu AU - Zhou, Xing-She AU - Li, Meng-Jie AU - Liu, Xin-Yue TI - OGU: Near-Optimal Group Selection of Heterogeneous Sensing UAVs via Capability Modeling and Aggregation T2 - Computer Modeling in Engineering \& Sciences PY - 2026 VL - 148 IS - 2 SN - 1526-1506 AB - Sensor-equipped Unmanned Aerial Vehicles (UAVs) are increasingly deployed for collaborative aerial sensing, yet selecting an optimal subgroup from a heterogeneous fleet remains challenging. Existing approaches rank individual UAVs by fixed, isolated metrics (e.g., sensor type, residual energy) and deploy them sequentially, failing to quantify task-specific performance under coupled operational uncertainties arising from platform heterogeneity, sensor configuration, and environmental dynamics. To address this, we propose Near-Optimal Group UAV Selection (OGU), a capability-driven modeling method. Rather than directly manipulating raw, heterogeneous hardware parameters, OGU aggregates each UAV–sensor unit into a capability entity characterized by intrinsic task-oriented attributes that are computed for the specific task context. This reformulation transforms group UAV selection into capability matching, reducing the optimization complexity from a high-dimensional coupled parameter space to a compact capability space. For an area coverage task, each heterogeneous UAV–sensor pair is abstracted as a unified Sweep Capability with two attributes—coverage benefit and energy cost—computed via sub-capability aggregation. An energy-efficient greedy heuristic algorithm then identifies a near-optimal UAV subgroup that fulfills task requirements with minimal total energy expenditure by leveraging these capability attributes. Empirically, the selected subgroup deviates from the exact MILP solution by less than a 5% optimality gap. Extensive simulations demonstrate that OGU outperforms the state-of-the-art selectivity-based Opti-U method and a random-selection baseline. OGU achieves a 131.3% higher energy efficiency and a 11.2% higher area coverage ratio while deploying fewer UAVs and reduces task execution time by approximately 7× through simultaneous group deployment. The proposed capability-oriented paradigm provides a flexible and scalable framework for managing heterogeneous sensing UAVs across diverse tasks. KW - UAV; sensor; capability aggregation; group UAV selection; energy efficiency; area coverage; random forest regressor DO - 10.32604/cmes.2026.085689