Unmanned aerial vehicle cluster dynamic coverage method based on multi-agent deep reinforcement learning

Abstract

The invention discloses an unmanned aerial vehicle cluster dynamic coverage method based on multi-agent deep reinforcement learning, and the method comprises the steps: firstly carrying out the modeling of a task through employing coverage energy and coverage power concepts; secondly, corresponding communication constraint force is set according to the requirement of communication keeping, a centralized action corrector is designed based on the corresponding communication constraint force, and the corrector only plays a role in training and is used for ensuring communication and generating positive and negative samples; and finally, the model is trained in cooperation with a multi-agent reinforcement learning algorithm, and a centralized action corrector is removed during execution, so that dynamic coverage of connectivity maintenance is realized.

Description

technical field [0001] The invention belongs to the technical field of dynamic coverage control of UAV swarms, and more particularly relates to a method for dynamic coverage of UAV swarms based on multi-agent deep reinforcement learning. Background technique [0002] In recent years, with the development of Internet of Things technology and communication technology, the application scenarios of drone swarms have become more and more extensive. UAV swarm refers to a multi-agent system composed of a group of UAVs with data acquisition and information processing functions. These drones can communicate to coordinate each other's actions to accomplish tasks that a single drone cannot. UAV swarms are increasingly used to perform dynamic coverage control tasks due to their high tolerance for high-risk and high-pollution environments. Typical dynamic coverage control tasks include aerial survey, target surveillance, disaster reconnaissance, etc., which have bright prospects and ou...

AI Technical Summary

Problems solved by technology

On the one hand, limited by the hardware platform and functional design, the communication range of UAVs is limited, and connectivity maintenance is inherently complicated; The nature preservation will limit the cluster expansion to maintain the communication connectivity, and the opposite and contradictory performances of the two dynamic behaviors make the cluster dynamic coverage task under the connectivity preservation constraint more complicated.

In the field of dynamic coverage, with the help of multi-agent deep reinforcement learning algorithms, existing methods can optimize the driving force of the model output towards the direction of maintaining cluster connectivity by imposing penalties on the driving force that causes the cluster to lose connectivity. The cluster connectivity during training cannot be guaranteed, and there is a problem of sparse rewards, and the model is not easy to converge

In view of the above deficiencies, the present invention proposes a UAV cluster dynamic coverage method based on multi-agent deep reinforcement learning. By introducing an action corrector, the driving force that causes the cluster to lose connectivity is corrected, which can ensure the connectivity of the cluster during training. , while solving the problem of sparse rewards

Images