Left ventricular assist devices (LVADs) technology requires developing and implementing intelligent control systems to optimize pump speed to achieve physiological metabolic demands for heart failure (HF) patients. This work aimed to design an advanced tracking control algorithm to drive an LVAD under different physiological conditions. The pole placement method, in conjunction with the sliding mode control approach (PP-SMC), was utilized to construct the proposed control method. In this design, the method was adopted to use neural networks to eliminate system uncertainties of disturbances. An elastance function was also developed and used as an input signal to mimic the physiological perfusion of HF patients. Two scenarios, ranging from rest to exercise, were introduced to evaluate the proposed technique. This technique used a lumped parameter model of the cardiovascular system (CVS) for this evaluation. The results demonstrated that the designed controller was robustly tracking the input signal in the presence of the system parameter variations of CVS. In both scenarios, the proposed method shows that the controller automatically drives the LVAD with a minimum flow of 1.7 L/min to prevent suction and 5.7 L/min to prevent over-perfusion.