The growing penetration of low-carbon technologies in residential networks such as photovoltaic generation (PV) units and electric vehicles (EVs) may cause technical issues on the grid. Thus, operation planning of electrical distribution networks (EDNs) should consider the inclusion of these technologies in order to avoid operational limit breaches. This paper proposes a dynamic scheduling method for the optimal operation of PV units and EVs in unbalanced residential EDNs, considering energy storage systems (ESSs). The proposed method optimizes the joint operation of PV units and EVs, using ESSs to increase the local consumption of the renewable energy. A rolling multi-period strategy based on a mixed integer linear programming model is used to dynamically optimize a centralized decision making, determining control actions for on-load tap changers (OLTCs), ESSs, PV units, and EVs connected to the network. At each time interval, data for PV generation and EV demand is updated using actual information and historical profiles, generating an updated forecast for a one-day-ahead operation in order to properly cope with weather uncertainties and EV owner's behavior without the need of multiple scenarios. The effectiveness and robustness of this approach are verified in different cases via a 107-node test EDN.