Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TempFuser: Learning Agile, Tactical, and Acrobatic Flight Maneuvers Using a Long Short-Term Temporal Fusion Transformer

Hyunki Seong, David Hyunchul Shim

Abstract

Dogfighting is a challenging scenario in aerial applications that requires a comprehensive understanding of both strategic maneuvers and the aerodynamics of agile aircraft. The aerial agent needs to not only understand tactically evolving maneuvers of fighter jets from a long-term perspective but also react to rapidly changing aerodynamics of aircraft from a short-term viewpoint. In this paper, we introduce TempFuser, a novel long short-term temporal fusion transformer architecture that can learn agile, tactical, and acrobatic flight maneuvers in complex dogfight problems. Our approach integrates two distinct temporal transition embeddings into a transformer-based network to comprehensively capture both the long-term tactics and short-term agility of aerial agents. By incorporating these perspectives, our policy network generates end-to-end flight commands that secure dominant positions over the long term and effectively outmaneuver agile opponents. After training in a high-fidelity flight simulator, our model successfully learns to execute strategic maneuvers, outperforming baseline policy models against various types of opponent aircraft. Notably, our model exhibits human-like acrobatic maneuvers even when facing adversaries with superior specifications, all without relying on prior knowledge. Moreover, it demonstrates robust pursuit performance in challenging supersonic and low-altitude situations. Demo videos are available at https://sites.google.com/view/tempfuser.

TempFuser: Learning Agile, Tactical, and Acrobatic Flight Maneuvers Using a Long Short-Term Temporal Fusion Transformer

Abstract

Dogfighting is a challenging scenario in aerial applications that requires a comprehensive understanding of both strategic maneuvers and the aerodynamics of agile aircraft. The aerial agent needs to not only understand tactically evolving maneuvers of fighter jets from a long-term perspective but also react to rapidly changing aerodynamics of aircraft from a short-term viewpoint. In this paper, we introduce TempFuser, a novel long short-term temporal fusion transformer architecture that can learn agile, tactical, and acrobatic flight maneuvers in complex dogfight problems. Our approach integrates two distinct temporal transition embeddings into a transformer-based network to comprehensively capture both the long-term tactics and short-term agility of aerial agents. By incorporating these perspectives, our policy network generates end-to-end flight commands that secure dominant positions over the long term and effectively outmaneuver agile opponents. After training in a high-fidelity flight simulator, our model successfully learns to execute strategic maneuvers, outperforming baseline policy models against various types of opponent aircraft. Notably, our model exhibits human-like acrobatic maneuvers even when facing adversaries with superior specifications, all without relying on prior knowledge. Moreover, it demonstrates robust pursuit performance in challenging supersonic and low-altitude situations. Demo videos are available at https://sites.google.com/view/tempfuser.
Paper Structure (16 sections, 9 equations, 7 figures, 2 tables, 1 algorithm)

This paper contains 16 sections, 9 equations, 7 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. Conventional Approaches
  4. Learning-based Approaches
  5. Flight Geometries
  6. Methodologies
  7. State and Action Representation
  8. Long Short-Term Temporal Fusion Transformer
  9. Soft Actor-Critic with State Trajectories
  10. Reward Function
  11. Experiment
  12. Environment Setup
  13. Baseline Schemes
  14. Evaluation
  15. Learned Flight Behavior
  16. ...and 1 more sections

Figures (7)

  • Figure 1: Dogfighting requires a combination of agility, tactics, and aerobatics to secure dominant positions over the opponent in complex airborne scenarios.
  • Figure 2: Human pilot's tactical maneuvers: pursuit strategies (left) and an out-of-plane maneuver, 'Low Yo-Yo' (right).
  • Figure 3: Geometries for aerial dogfights.
  • Figure 4: Overview of the TempFuser architecture for the policy network with the following example sets $n_s = n_l = \Delta t = 4$.
  • Figure 5: The pursuit score function and the altitude reward term.
  • ...and 2 more figures