The Quadcopter or Quadrotor Helicopter is becoming an increasingly popular aircraft for both personal and professional use. Its maneuverability lends itself to many applications, from last-mile delivery to cinematography, from acrobatics to search-and-rescue.
But it also comes at a price–the high complexity of controlling such an aircraft makes it almost impossible to manually control each individual motor's thrust. So, most commercial quadcopters try to simplify the flying controls by accepting a single thrust magnitude and yaw/pitch/roll controls, making it much more intuitive and fun.
This project enables quadcopters to autonomously achieve desired control behaviors such as takeoff and landing using reinforcement learning.
