Teaching Machines to Fly Themselves using Artificial Intelligence
Autonomous flight has many challenges and the stakes involved are high. This hasn’t stopped many people from working on autonomous flight.
Because of its complexity and the human lives involves, commercial air travel will likely remain the domain of human pilots for the near future. However, there are still many important applications for self-piloted air vehicles.
The most immediate ones that come to mind are small drones. These have a huge variety of important applications, ranging from agriculture to humanitarian work and logistics. For example, Zipline operates in Rwanda and can rapidly transport blood products and other vital medical supplies throughout the country.
Zipline’s drones and many others are not piloted by a human sitting at home or in an office. They are flown reliably using algorithms. These algorithms follow a set of rules and flight plans, but teaching algorithms exactly how to manipulate the control surfaces to achieve a desired maneuver is something increasingly achieved using machine learning.
Writing a basic autopilot in Python using reinforcement learning
Suppose you wanted to teach a machine to fly a basic maneuver, how would you even start this process?
