This paper addresses the decentralized task allocation problem of multirobot systems, in which the objective is to maximize the total task assignments, i.e., the number of tasks that can be successfully executed by all robotic vehicles under the time constraints of tasks and battery limits of vehicles. Based on the state-of-the-art performance impact (PI) algorithm, a novel extension named PI for minimizing traveling time (PI-minTravel) is proposed in this paper. With the proposed PI-minTravel, tasks that are close enough to the last task of each vehicle are assigned to the vehicle first, so that the total traveling time of all vehicles can be minimized. Due to the limited fuel of each vehicle, less traveling time will leave more time to execute tasks, then more tasks can be executed, especially when the ratio of tasks to vehicles is high. Extensive simulation results show that the proposed PI-minTravel can assign more tasks and converge within fewer iterations compared with PI algorithm, while it can assign fewer tasks but converge within much fewer iterations compared with PI for maximizing assignments (PI-maxAss) algorithm.