| To transform coarse-grained reconnaissance requirements into autonomous bidding inputs and separate execution feedback from observation conclusions, a reconnaissance task order generation mechanism was proposed for multi-UAV reconnaissance in unknown urban areas. Connected road clusters were used as basic spatial units, with staged release, execution–observation separation, road-state updating, and event-triggered incremental order generation, while preserving the standard CBBA procedure. In S1, the peak CBBA input decreased from 50 to 10, erroneous negative evidence from 22.77 to 0, and event–order traceability reached 100%. In S5 with 500 road clusters and 50 UAVs, the median computation time of each order-processing stage was below 10.11 ms, demonstrating efficient dynamic order generation. |
