A Satellite Derotation Method Based on Deep Reinforcement Learning

Abstract

The invention discloses a satellite derotation method based on deep reinforcement learning, which is characterized in that it comprises the following steps: marking data samples of known satellites to establish a sample data set of known satellites; using a fully convolutional neural network to train the sample data set , so that the terminal can understand and identify the known satellites in the image or video, and obtain the confidence map of the key points of the known satellites in the image or video; track the trajectory of the key points in the video, and estimate the position of the known satellite through the PNP algorithm Attitude; the optimal derotation is trained by the DDPG algorithm, and the derotation of the space manipulator brushes the side of the spacecraft sail to complete the satellite derotation. The method of the present invention realizes the derotation of the out-of-control satellite with high-speed spin by means of deep reinforcement learning, and at the same time combines the visual information to allow the computer to contact the data and model environment, train the optimal grasping pose, and improve the accuracy of the target capture of the space manipulator Spend.

Description

technical field [0001] The invention relates to a satellite derotation method based on deep reinforcement learning, and belongs to the technical field of satellite derotation methods. Background technique [0002] With the increase in the number of spacecraft in orbit and their wide application, real life is increasingly inseparable from the various application functions provided by spacecraft in orbit. Due to the limitation of the space on-orbit working organization's own conditions and the influence of the space environment, without any supply and maintenance, the operation is often forced to stop due to limited fuel, outdated equipment or module failure, and a new system has to be remanufactured and launched to replace it. Replacement, resulting in unnecessary losses and waste. GEO is the geosynchronous orbit. Carrying out GEO on-orbit maintenance and service and research on related technologies can effectively prolong the service life of the on-orbit system, and at the ...

AI Technical Summary

Problems solved by technology

Q-learning makes it possible to find the optimal action strategy without the knowledge of immediate reward function and state transition function. In other words, Q-learning makes reinforcement learning no longer dependent on the problem model, but still needs to know the final reward or goal state

Images