Modern urbanization is experiencing an increased number of autonomous vehicles day by day leading to traffic congestion, parking problems, and overcrowding at the parking lots are one of the major concerns that the smart cities are facing nowadays and due to floating prices of inflammable gases recently there is a drastic escalation of electric motor vehicles (EMV) on the roads needs Parking with charging facility. This research explores and resolves the issues of dynamic provisioning of parking spaces for EMV with or without charging facility in three phases, firstly using cluster-based searching for vacant parking slots, secondly using deep reinforcement learning (DRL) based user parking request processing using cloud server and distributed fog nodes with the assistance of Vehicular Adhoc networks (VANET) model and finally using evolutionary optimization based social ski driven routing (SSD) algorithm is used to efficiently route the EMV to the vacant parking lots. The fitness function is newly devised considering multi-objective parameters such as traffic density,battery power As well as distance parameter and shortest routing decision are made, Hence by using proposed cluster-based parking system using deep reinforcement learning-based Social Ski driven (DRL-SSD) routing protocol achieves allocation of parking slots with a high probability of success rate of 95% with minimal traffic density 7.5 per lane, minimal delay of 7.55min and minimal fitness value of 15.88.