We propose a reinforcement learning algorithm to learn an adaptive flight speed policy tailored to varying environment complexities, enhancing obstacle avoidance safety. A downside of learning-based ...