This thesis addresses the problem of autonomous landing of a hybrid quadplane UAV under localized turbulent wind conditions. The UAV is modeled in hover configuration, where only the upward facing rotors are used for control allocation. A Proportional–Integral–Derivative (PID) controller is implemented as a low-level tracking controller. As a baseline, a finite-state machine (FSM)-based guidance law comprising approach, hover, and descent phases with predefined transition criteria is benchmarked against a reinforcement learning (RL)-based guidance policy. The RL agent generates velocity commands that are tracked by the PID controller. Simulation results demonstrate that the RL-based policy leverages real-time wind disturbance observations to adapt its trajectory, avoiding hazardous downwash regions. The agent additionally learns the typical geometric position and size of the localized wind field encountered during training, and incorporates this as a prior in its guidance strategy alongside its real-time sensing capability. Under a randomized wind field around the landing platform, the proposed approach achieves a landing success rate of 66%, compared to 24% for the baseline. The RL-based method additionally reduces both overall landing time and touchdown velocity. ... Read more