This article proposes a robust resilience enhancement method for a power distribution network in ice storms by the optimal routing of mobile de-icing devices (MDIDs) on congested transportation roads. We consider MDIDs as emergency vehicles, as other vehicles yield the right-of-way to MDIDs, and propose a de-icing schedule (DIS) to illustrate how MDIDs are routed on congested transportation roads. We further coordinate the DIS and MDID routing, which mitigate transportation routing congestions, with power distribution system operation, which considers distribution network reconfiguration and distributed energy resource dispatch. A two-stage robust model is proposed to manage the effects of ice storms forecast errors on both power distribution and urban transportation networks. The proposed model is reformulated as a mixed-integer second-order cone programming problem. The Benders decomposition and column-and-constraint generation algorithms are further utilized to solve the proposed MISOCP. Numerical results for the modified IEEE 33-bus 12-node, IEEE 123-bus 25-node, and 252-bus 80-node electricity-transportation systems show the effectiveness of the proposed model and solution technique for enhancing the power system resilience.