Unmanned aerial vehicle (UAV) nowadays are inexpensive and can serve as a robust communication platform in the sky. Using UAV as access point can be particularly helpful for wireless terminals in areas where terrestrial infrastructure is absent or damaged. The problem is how to deploy a fleet of UAVs to efficiently cover scattered terminals while not wasting too much energy on the deployment process. In this paper, we propose two autonomous service deployment approaches for UAVs based on game theory. In these approaches, UAVs adaptively adjust their locations based on local information rather than instructions from a control station. Experimental results show that the proposed approaches outperform existing approach in terms of average spectral efficiency and UAV travelling distance.