Online distributed mission planning method of heterogeneous multi-UAV collaborative mapping

This paper addressed the decision-making challenges faced by heterogeneous multi-UAV cooperative mapping systems in dynamic environments. The research focused on constraints imposed by moving targets and obstacles during multi-UAV mapping operations. By integrating pre-planned flight routes, a critical element in cooperative mapping scenarios, the approach overcomes the limitations of traditional distributed auction planning methods, namely local optima and task conflicts. Ultimately, this method significantly enhances the system′s real-time responsiveness and task execution efficiency in unknown environments.

This paper proposed an improved hierarchical distributed task planning framework that integrates the advantages of pre-planning and dynamic planning: The preprocessing layer defined flight path attributes based on pre-planned trajectories, constructing a task valuation model that incorporates additional flight time and global planning time to achieve comprehensive cost consideration. The distributed task allocation layer designed a local auction algorithm under constrained communication, resolving task conflicts and ensuring decision consistency through synchronized bidding, auctioning, and market pricing among UAVs. The post-processing layer employed an extended Dubins 3D model for trajectory smoothing updates. Additionally, a joint trajectory correction algorithm based on rolling-horizon predictive control was proposed. This algorithm constructed a benefit function encompassing area coverage, obstacle avoidance, and maneuver reduction, ensuring flight safety through path prediction and trajectory correction.

Compared to traditional market auction methods, the proposed algorithm effectively avoids "myopic" decision-making by incorporating pre-planned global information, which resulted in average task completion time reductions of 8.4%, 24.2%, and 29.1% across three simulation scenarios. The joint trajectory correction strategy increased the minimum distance between UAVs and obstacles from 35.9 m to 122.7 m, meeting safety radius requirements. Across multi-scenario simulations, the UAVs achieved 100% area coverage while dynamically responding to unexpected events such as target detection and obstacle identification. Task assignments remained conflict-free and aligned with the global efficiency optimization objective.

The proposed distributed planning method combining pre-planning with dynamic planning demonstrates excellent applicability and reliability. Its hierarchical framework, task valuation mechanism, local auction algorithm, and integrated trajectory correction strategy simultaneously meet the coverage and efficiency requirements of static surveying while adapting to the complex constraints of dynamic environments. This approach provides an efficient and feasible solution for heterogeneous multi-UAV collaborative emergency surveying.