Dynamical mobile task allocation, by which tasks can move randomly before they are assigned robots to execute. For such a new task assignment domain, we propose a hybrid dynamic mobile task allocation and reallocation method that combines our previous proposed dynamical sequential method and global optimal method. Robots bid for tasks and transmit the costs to other robots. Then all robots select tasks from the combinatorial cost table to minimize the objective function. During the next time step, robots continue to select the assigned tasks for which costs are smaller than the set thresholds. Alternatively, robots for which costs exceed the corresponding threshold rebid unassigned tasks and transmit the calculated costs to others. The un-selected robots then re-select unassigned tasks from the combinatorial cost table according to global optimal task allocation method. In this study, the advantages of the proposed approach are demonstrated by comparison with existing task allocation methods. The simulation results demonstrate that a system implementing our method can obtain maximal accomplished efficiency of whole system and minimal executed costs for each individual robot. The negotiation time steps, communication costs and computational times are reduced using the proposed algorithm. Moreover, we believe that our method can extend the previous methods to be suitable for a large-scale distributed multi-robot coordination system.