Method for Optimizing an Integrated Flight Control System using Reinforcement Meta-Learning

This technology is a method to incorporate reinforcement meta-learning to the design of a flight control system for missiles. The integration will benefit the control system, as it optimizes it with the help of a simulation environment. The technology removes the need for manual tuning of the control system settings.

Background: 
The flight control system on a missile is the key element that enables the missile to hit the system performance requirements. Its main purpose is to force the missile to achieve the maneuvering commands that are input by the guidance system. The settings of the control system need to be calibrated beforehand to achieve the desired performance requirements. 

One way to optimize and resolve the issue of manually tuning the control system settings is to use reinforced meta-learning. Reinforced meta-learning is a branch of machine learning that aims to adapt models in a quick manner, to perform new tasks by learning the underlying structure that is found across related tasks. This enables the control system to have better settings than by a standard manual tuning.

Applications: 

  • Flight control systems for missiles


Advantages: 

  • Removes the need for manual tuning of the control system
  • Better tuning settings due to the use of the simulated environment
Patent Information: