Heterogeneous Cooperative Task Planning for Time-Constrained Multi-Object Response Under Incomplete Information

Abstract

This paper studies heterogeneous cooperative task planning for time-constrained multi-object response under incomplete information. The problem is motivated by dense dynamic-object scenarios in which multiple resource types must be coordinated within limited response windows while both object states and object attributes are imperfectly known. First, estimation errors in object position, speed, heading, trajectory, and arrival time, together with recognition biases in object type, intent, and composite priority level, are incorporated into a unified dual-uncertainty representation. Second, based on threshold-based cumulative response effects, a package-based planning model is established by considering resource heterogeneity, platform concurrency, consumable capacity, and cooperative cumulative effects. Third, an improved adaptive large neighborhood search framework is designed to match the proposed model and support planning in scenarios of different scales. Finally, computational experiments are conducted from both model-validation and algorithm-validation perspectives. The results show that dual-uncertainty modeling and heterogeneous cooperation improve out-of-sample scheme quality, worst-case scheme quality, and critical-object handling performance. The proposed method is close to exact branch-and-bound on ultra-small instances and clearly outperforms a greedy baseline on small-, medium-, and large-scale instances.