敏捷卫星同时具备滚转与俯仰机动能力，可采用条带拼幅模式对区域目标观测成像，有助于提升区域目标覆盖效率。在该多条带观测调度过程中，需同时优化敏捷卫星的滚转角、俯仰角及各条带的观测起止时间，导致调度问题复杂度激增，本文提出基于网格剖分的敏捷卫星区域目标观测调度方法。首先，采用六边形网格对目标区域进行双层离散，用于覆盖度量、姿态映射与启发式引导。其次，以覆盖率最大化为目标，建立考虑多约束的混合整数非线性规划模型。为实现模型求解，设计基于自适应反馈迭代的双层优化框架，并提出融合模拟退火与协同粒子群遗传的双层优化算法。其中上层通过模拟退火及其多邻域策略进行观测条带分配；下层采用协同粒子群遗传算法进行俯仰角和观测窗口的规划。大量对比实验表明，在敏捷卫星区域目标观测场景中，所提出方法相较不同上、下层对比算法能够获得更高的覆盖率与更好的收敛稳定性。通过与传统对地观测模式的对比可以发现，敏捷卫星的条带拼幅模式能够显著提升区域覆盖率。

Agile satellites possess both roll and pitch maneuverability and can employ a strip-mosaicking observation mode for regional target imaging, thereby improving coverage efficiency. In this multiple-strip observation scheduling process, the roll angle, pitch angle, and the start and end times of each observation strip must be jointly optimized, which greatly increases problem complexity. To address this, we propose a grid-tessellation-based scheduling method for regional target observation with agile satellites. First, a hexagonal grid is used to perform two-level discretization of the target area for coverage quantification, attitude mapping, and heuristic guidance. Second, a nonlinear mixed-integer programming model is formulated to maximize coverage under multiple constraints. To solve the model, we design a bilevel optimization framework with adaptive feedback iteration and develop a bilevel algorithm that integrates simulated annealing with a cooperative particle swarm–genetic method. The upper level allocates observation strips via simulated annealing with multi-neighborhood strategies, while the lower level plans pitch angles and observation windows using the coopera-tive particle swarm–genetic algorithm. Extensive comparative experiments show that, in agile-satellite regional observation scenarios, the proposed approach achieves higher coverage and more stable convergence than alternative upper- and lower-level baselines. Fur-thermore, compared with traditional Earth observation modes, the strip-mosaicking strategy significantly increases regional coverage.