Network overloading and electric power shortfalls occur due to large variations in anticipated renewable electric power and electric loads, which may cause system overheating or network failure. A large portion of residential loads (RLs) comprises of electric vehicles (EVs); therefore, by using in-system electric vehicle batteries (EVBs), above mentioned issues can be resolved. A lot of research has been done in this domain but the stochastic nature of this system has not been fully explored yet. In this article, an intelligent charging mechanism of EVs has been proposed for RLs. EVBs are charged during off-peak hours and then batteries are utilized as an independent power source during peak hours to decrease stress on the power network. Vehicle-to-grid charging and discharging scenarios are considered, to optimize the rate of charging. A novel stochastic model predictive control (SMPC) framework, based on smart charging/discharging of EVBs has been proposed in this article. SMPC deals with the uncertain variations in the projected generation and demand of electric power, due to variable renewable energy and high electric loads, respectively. It deals with the discrepancy of supply, demand, and energy prices. Using SMPC, complex stochastic problems can be transformed into a simpler optimization problem, making it computation efficient and suitable for real-time applications. To relax constraints on power demand, a worst case scenario technique has been used which reduces the computational cost of the system. Simulation results depict that controlled charging and discharging of EVBs in power grid network help reduce overshoots and undershoots of electric power demand in a smart community. Comparison of the proposed strategy has been done with benchmark algorithms, to show the competency of the proposed strategy. This study provides insight of a smart community microgrid operation and resultant trade-offs to manage the demand and supply of electricity.