Utilizing the Python programming language and the Tkinter library, I propose to develop a simulated robotic application tailored for search and rescue missions. Central to this endeavor is the integration of a neural network, trained on pertinent search and rescue data, to effectively guide the robotic agent towards areas with a high probability of locating survivors. The neural network's computations yield a dynamic cost map, which takes into account both traversal costs and the priority of various regions within the search area. This cost map serves as a comprehensive tool, continuously furnishing the robotic agent with insights into the search region and optimizing its search strategy. Importantly, this approach is conceived to address challenges inherent in reinforcement learning, such as the lack of a global view and the need to balance multiple objectives in search and rescue mission.