The energy storage system (ESS), as well as its efficient management, represents key factors for the success of Electric Vehicles (EVs). Due to well-known technological constraints in the ESS, there has been a growing interest in combining various types of energy sources with complementary features. Among the many possible combinations, our interest here lies in the hybridization of batteries and supercapacitors (SCs), with an active parallel arrangement, i.e., the sources are connected to the DC bus through two bidirectional DC-DC converters (step-up). Based on this ESS topology, a robust DC-Link controller is employed to regulate the DC-bus voltage and track the SCs current, in spite of uncertainties in the system. For this purpose, we start by showing that the converters uncertainty, e.g., the powertrain load, can be modelled as a convex polytope. The DC-Link controller is then posed as a robust Linear-quadratic Regulator (LQR) problem and, by exploring the convex polytope, converted in a Linear Matrix Inequalities (LMI) framework, which can be efficiently solved by numerical means. Finally, the operation envelope of the controller is extended by scheduling the gains according to energy sources voltages, an important feature to cope with the voltage variations in the SCs. To analyze the performance of the control architecture, a reduced-scale prototype was built. The experimental results show that, compared with the non-robust and non-gain-scheduled controllers, the proposed DC-Link controller offers a better transient response and robustness to disturbances. Further, the global performance of the controller is also evaluated during some driving cycles.