Automated driving in urban settings is challenging. Human participant\nbehavior is difficult to model, and conventional, rule-based Automated Driving\nSystems (ADSs) tend to fail when they face unmodeled dynamics. On the other\nhand, the more recent, end-to-end Deep Reinforcement Learning (DRL) based\nmodel-free ADSs have shown promising results. However, pure learning-based\napproaches lack the hard-coded safety measures of model-based controllers. Here\nwe propose a hybrid approach for integrating a path planning pipe into a vision\nbased DRL framework to alleviate the shortcomings of both worlds. In summary,\nthe DRL agent is trained to follow the path planner's waypoints as close as\npossible. The agent learns this policy by interacting with the environment. The\nreward function contains two major terms: the penalty of straying away from the\npath planner and the penalty of having a collision. The latter has precedence\nin the form of having a significantly greater numerical value. Experimental\nresults show that the proposed method can plan its path and navigate between\nrandomly chosen origin-destination points in CARLA, a dynamic urban simulation\nenvironment. Our code is open-source and available online.\n