
3827. Dynamic Mass-Aware Trajectory Tracking of Airships Using Multi-Actors Proximal Policy Optimization
$20.00
SAWE Members get a $200 store credit each year.*
*Store credit coupon available at checkout, click the button in your shopping cart to apply the coupon.
Not applicable to SAWE textbooks and current conference technical papers.
Paper
Multi-Actors Proximal Policy Optimization. 2025.
Abstract
Dynamic mass variations significantly influence the attitude and trajectory tracking performance of stratospheric airships. To address this challenge, this paper proposes a dynamic
mass-aware control algorithm for airships using Multi-Actors Proximal Policy Optimization (PPO), a deep reinforcement learning framework. We first establish a comprehensive airship dynamics model that explicitly accounts for varying mass characteristics, formulating the state space, action space, and reward function to capture the impact of payload shifts or fuel consumption on flight stability. Multi-Actors PPO, leveraging a clipped probability ratio objective, enhances policy update stability and data efficiency in the presence of mass disturbances. Neural networks are employed to approximate the policy and value functions, while Generalized Advantage Estimation (GAE) further boosts optimization performance. Preliminary analyses under diverse flight conditions and dynamic mass scenarios suggest that the proposed approach can significantly outperform traditional controllers such as PID and LQR in terms of trajectory tracking accuracy and robustness. Consequently, it offers an effective and stable solution for dynamic mass-aware intelligent control in unmanned airship systems.