A local perception based dynamic task scheduling algorithm (LPDTAA) is proposed to address the dynamic task allocation problem for distributed Unmanned Aerial Vehicle (UAV) clusters. First, a local auction strategy is proposed to adapt to dynamic goals based on the CBBA; second, a task set reordering strategy is designed to eliminate task conflicts generated in local auctions globally; finally, two information consistency algorithms are proposed to reduce the number of task conflict sequence releases and speed up the convergence. Simulation experiments show that the total cost of tasks under initial allocation and reallocation of the algorithm is lower than that of the CBBA, and close to the Particle Swarm Optimization (PSO) algorithm in the classical centralized algorithm, the running time of initial allocation and reallocation is smaller than that of CBBA and much smaller than the PSO.