Objective The status of key areas outside urban rail transit stations is an important basis for implementing passenger flow control and synergistic organization inside and outside the station during the initial stage of emergency response to sudden large passenger flows. Limited by the coverage of fixed monitoring and the efficiency of manual inspections, it is difficult to obtain status information of dispersed areas in a timely manner. Therefore, there is an urgent need to study low-altitude sensing task scheduling methods under the condition of limited UAV （unmanned aerial vehicle） resources.

Method Based on the spatial functions and risk differences outside the station, a multi-UAV low-altitude sensing task scheduling model is constructed, coordinating the risk-weighted status acquisition time and the UAV flight time consumption. Comprehensively considering constraints such as task allocation, path continuity, sensing sequence, and endurance, the RT-ALNS （risk-time adaptive large neighborhood search） algorithm is adopted to solve the model. A calculation example is constructed for verification based on the dispersal scenario of a large-scale event at a metro station in Tianjin.

Result & Conclusion  Compared with the shortest distance scheme and the risk priority scheme, the risk-weighted status acquisition time of the RT-ALNS scheduling scheme is reduced by 21.70% and 5.26%, respectively. Its optimal objective value is consistent with that of the GA （genetic algorithm） and superior to that of the SA （simulated annealing） algorithm. The mean, standard deviation, worst value, and RPD （relative percentage deviation） of the objective function for the RT-ALNS algorithm are all lower, with an average solution time of 4.51 s. The proposed multi-UAV low-altitude sensing task scheduling model and RT-ALNS scheduling scheme can coordinate the sensing timeliness in high-risk areas and UAV flight time consumption, which can support rapid sensing and emergency organization work outside the station in sudden large passenger flow scenarios.