The ubiquitous application of autonomous drones across various industries, such as logistics, surveillance, agriculture, and disaster response, has necessitated smart and responsive fleet management systems. In this paper, a Markov Decision Process (MDP)-based framework for optimal task allocation and route planning in a fleet of autonomous drones operating in dynamic and uncertain settings is developed. The model captures the state of each drone as a function of its location, battery level, and task status, and transitions as consisting of task selection, direction, and recharging action. Transition probabilities consider external uncertainties such as wind speeds and moving obstacles, and the reward function seeks to maximize the effectiveness of task attainment, minimize energy consumption, and finish the mission on time. By solving the MDP through value iteration and policy iteration algorithms, the system allows real-time decision-making, optimizing operational performance in accordance with system limitations. Experimental simulations demonstrate that the MDP strategy significantly enhances overall mission efficiency and robustness relative to greedy and heuristic methods, particularly under stochastic and resource-constrained conditions. The framework forms the basis for scalable and robust management of drone fleets in increasingly demanding operational environments.