A Rust-Implemented Dueling Double DQN for Fixed-Wing Autonomous Flight
Saugat Chaudhary Tharu *
Department of Computer Science, (SOS), Noida International University, Greater Noida, India and Department of Computer Science, IIMS College, Kathmandu, Nepal.
Shrutika Ojha
Department of Computer Science, (SOS), Noida International University, Greater Noida, India and Department of Computer Science, IIMS College, Kathmandu, Nepal.
Rija Bhomi
Department of Computer Science, (SOS), Noida International University, Greater Noida, India and Department of Computer Science, IIMS College, Kathmandu, Nepal.
Dipen Gurung
Department of Computer Science, (SOS), Noida International University, Greater Noida, India and Department of Computer Science, IIMS College, Kathmandu, Nepal.
Sambid Shrestha
Department of Computer Science, (SOS), Noida International University, Greater Noida, India and Department of Computer Science, IIMS College, Kathmandu, Nepal.
Anmol Adhikari
Department of Computer Science, (SOS), Noida International University, Greater Noida, India and Department of Computer Science, IIMS College, Kathmandu, Nepal.
*Author to whom correspondence should be addressed.
Abstract
Deep reinforcement learning for autonomous unmanned aerial vehicle control has largely been demonstrated with multirotor platforms and high-level machine-learning frameworks. This study presents a Dueling Double Deep Q-Network (D3QN) training pipeline implemented in Rust without an external machine-learning library and integrated with Godot 4 through GDExtension for fixed wing flight control. The controller addresses fixed-wing requirements, including airspeed maintenance, lift management, throttle regulation, stall avoidance, and coordinated turning. The network combines a duelling architecture, double Q-learning, prioritised experience replay, and three-step returns in a 512→256 hidden-layer configuration containing 142,088 parameters for the 16-dimensional input case. The agent selects among seven discrete actions and supports both 12-dimensional and 16-dimensional observation spaces through a cross-dimensional weight-transfer procedure. Training was conducted for 200 episodes using four random seeds. Across seeds, the mean best episodic reward was 4825±40, while the coefficient of variation for best reward was 0.8%. In the final 30 episodes, no crashes were recorded, although completion rates varied substantially between seeds.
Airspeed remained within ±8 m/s of the 50 m/s target. Batch-64 gradient updates required less than 1 ms, representing an approximately 35-fold reduction in latency relative to the preceding GDScript implementation, and the reported runtime memory footprint remained below 50 MB. These findings support the feasibility of native Rust-based D3QN training for real-time fixed-wing simulation, while the observed inter-seed variability indicates that reward shaping and convergence robustness require further evaluation.
Keywords: Autonomous fixed-wing UAV, deep reinforcement learning, duelling double deep Q-network, Rust, Godot 4, GDExtension, prioritised experience replay, fixed-wing flight control, cross-dimensional state space, real-time training