Unmanned aerial vehicle maneuvering avoidance decision-making method based on deep reinforcement learning

Abstract

The invention provides an unmanned aerial vehicle maneuvering avoidance decision-making method based on deep reinforcement learning, and belongs to the technical field of unmanned aerial vehicles. According to the technical scheme, the method comprises the following steps: S1, acquiring friend or foe situation information according to an airborne sensor system of an unmanned aerial vehicle; s2, constructing an unmanned aerial vehicle maneuvering avoidance decision deep reinforcement learning model structure; s3, constructing a hierarchical target-oriented learning model structure; s4, completing unmanned aerial vehicle maneuvering avoidance decision-making method learning according to interactive training; and S5, deploying and applying the unmanned aerial vehicle maneuvering avoidance decision-making method. The method has the advantages that the unmanned aerial vehicle can be endowed with the learning ability from shallow to deep, the unmanned aerial vehicle can autonomously complete maneuvering avoidance decisions, and the survival ability of the unmanned aerial vehicle on a battlefield is improved.

Description

technical field [0001] The invention relates to the technical field of unmanned aerial vehicles, in particular to a decision-making method for maneuvering and avoiding unmanned aerial vehicles based on deep reinforcement learning. Background technique [0002] At present, under the background of airspace integrated intelligent combat, UAV technology has been widely used in military operations. It has achieved great success in many fields, such as cooperative reconnaissance, ground attack, destroying the enemy's air defense system, etc. In the application process, effectively avoiding threats and improving the survivability of UAVs in dynamic confrontation environments is the key to the success of combat missions. With the complexity and changeability of the modern battlefield environment, it is becoming more and more difficult for operators to complete complex flight tasks through manual operation. The survivability and combat capability of UAVs play a vital role in milita...

AI Technical Summary

Problems solved by technology

But designing reward functions is challenging for more complex tasks that require domain-specific knowledge

In addition, reward shaping may bias the policy toward undesirable behavior and limit the agent's access to safe actions

Images