Reinforcement learning-based distributed optimal formation tracking control with obstacle avoidance for quadrotor UAVs | Synapse