Improved Aircraft Trajectory Language Planning Strategy Based on Soft Actor Critic

Authors

  • Yuncheng Dong, Yupeng Wang

Keywords:

Trajectory planning algorithm, Artificial Potential Field, Soft Actor-Critic

Abstract

This paper introduces an innovative trajectory planning algorithm for stratospheric airships, combining Artificial Potential Field (APF) method with Soft Actor-Critic (SAC) deep reinforcement  learning  strategy,  aimed  at  enhancing  navigation  performance  of airships  in complex near-zero wind layer environments. The algorithm design encompasses a state space including wind field data and airship  status, with decision-making  facilitated by a custom- designed five-layer network architecture. Notably, the reward function comprehensively takes into account target proximity, no-fly zone avoidance, and energy management, ensuring both efficiency and safety in flight operations. Experimental results indicate that the proposed APF- SAC algorithm, compared to conventional SAC, achieves faster convergence, higher cumulative rewards, fewer action steps, and a 97% navigation success rate in simulated environments.

Downloads

Published

2024-06-11

How to Cite

Yuncheng Dong, Yupeng Wang. (2024). Improved Aircraft Trajectory Language Planning Strategy Based on Soft Actor Critic. Onomázein, (64 (2024): June), 70–78. Retrieved from http://www.onomazein.com/index.php/onom/article/view/682

Issue

Section

Articles