This study mainly focuses on comparison of multi-robot task allocation problem and evaluation of corresponding solutions. Auction has been used in the problem widely. Various kinds of auction have been proposed previously. In this paper, we compare three auction-based algorithms applied in the task allocation problem from different aspects. In addition, an incremental variant is adopted in the parallel auction to improve its performance. We also compare the improved auction with the general parallel auction. Simulation results demonstrate that the improved parallel auction achieves better performance on the overall cost and completion time than the general parallel auction. Moreover, we summarize the suitable conditions for the algorithms we mentioned in the paper based on the simulation results.