High-accuracy localization and formation are essential for multi-UAV networks to perform cooperative tasks. However, the joint design of localization and motion planning is challenging due to complex information coupling effects, which leads to a loss of formation accuracy. In this paper, we establish an integrated localization and motion planning scheme for multi-UAV networks. First, we derive bounds for the relative formation error, which reveals how measurement and motion noises affect the formation accuracy. Then, we propose a bidirectional process framework to enhance the formation accuracy. The forward process presents a near-optimal motion planning algorithm that leverages the equivalence relation of relative formations to mitigate the impact of localization uncertainties. The backward process addresses bandwidth allocation and UAV activation to maximize formation accuracy. Numerical results verify the gains of the proposed integrated scheme in formation accuracy.