Organizations: State Key Laboratory of ISN and School of Telecommunications Engineering, Xidian University, Xi’an 710071, China · School of Aerospace Science and Technology, Xidian University, Xi’an 710071, China · Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, N2L 3G1, Canada
Autonomous aerial vehicles (AAVs) enable data collection for sixth-generation Internet-of-Things networks, but their trajectories couple nonlinear wireless rates with long-horizon service progress. This paper views the evolution of AAV kinematics, channel state, and user backlog as a structured differentiable world model and develops Learn for Variation (L4V) to exploit that model efficiently. L4V replaces a discontinuous completion-time objective with a cumulative-backlog surrogate, unrolls the mission dynamics, and propagates pathwise sensitivities to a neural policy through the discrete adjoint recursion. The resulting derivative is exact conditional on a fixed exogenous-noise realization; stochastic expected-objective optimization still requires sampling. We show that the structured adjoint grows at most polynomially with the horizon and establish a stationary-point rate for fixed-step full-gradient descent under standard smoothness assumptions. The framework also learns shared OFDMA allocation under reparameterized shadowing and Rician fading, while distributional pretraining amortizes model-based optimization into forward-only deployment on unseen layouts. Paired stress tests cover channel-generator mismatch, noisy partial observations, a fixed-resource two-AAV extension, and a circular no-fly region. Code and configurations are available at https://github.com/UNIC-Lab/L4V-AAV. Against genetic-algorithm, DQN, A2C, DDPG, and differentiable model-predictive-control implementations, L4V reduces mission time by up to 65%, executes a default mission in 53 ms, and completes all 60 frozen-policy tests after pretraining on 1,600 layouts.
Figures & tables
Fig. 1 : The L4V framework. Each blue transition block is one step of the structured wireless world model. The forward pass (black) unrolls the AAV trajectory and accumulates L ; the backward pass (red) implements the discrete adjoint and propagates exact conditional pathwise gradients ∇θL to the policy parameters.
Fig. 2 : Performance versus service-area size L (a, b) and noise power σ2 (c, d). Medians over 10 trials with inter-quartile bands; completion time on a log axis.
Fig. 3 : Performance versus user density K (a, b) and channel gain η (c, d). Medians over 10 trials with inter-quartile bands; completion time on a log axis.
Method
Scenario optimization
Execution/solve
L4V
293 s
53 ms
GA
–
2369 s
DQN
887 s
118 ms
A2C
2475 s
131 ms
DDPG
3506 s
111 ms
MPC
–
64 s
TABLE I : Scenario-specific optimization and execution costs at the default configuration. Neural execution is the warmed median rollout cost scaled to the observed median mission length; GA and MPC report median per-instance solve time. Held-out transfer is evaluated separately in Fig. 6(e) .
Fig. 4 : Allocation ablation and energy-weight sensitivity. Panels (a, b) compare learned allocation with equal split over five paired instances; thin segments preserve each pairing and thick horizontal bars mark medians. The broken axes omit empty ranges while retaining and labeling the capped run. Panel (c) shows every energy-sweep trial with median/IQR summaries and a linear diagnostic. Panel (d) reports paired percentage changes from γ=0 with inter-quartile bands. Lower is better throughout, and capped or regressing trials are retained.
Fig. 5 : Executed AAV paths on one shared K=4 instance; marker size scales with task demand and each path is labeled with its mission length.
Fig. 6 : Reviewer-driven stress tests. Panels (a, b) keep the trained policy fixed and pair the random draws across deployment conditions. The nominal channel is (κ,Kr,ν)=(0.15,6,2) ; mild and severe shifts are (0.25,3,2.2) and (0.40,0,2.5) . Observation tests add a fixed location bias of standard deviation 0.30 , independently mask each current backlog update with probability 0.50 while holding its last estimate, or combine both. Panel (c) retains one global radio budget as the fleet grows from one to two AAVs. Panel (d) applies hard collision checks after training with or without obstacle geometry and a soft clearance barrier. Panel (e) freezes one distributionally pretrained policy on disjoint layouts. Panel (f) reports completion and circular turn variation. Thin marks retain trial-level outcomes; thick marks and bands show medians and IQRs; capped runs and success rates are retained. Logarithmic completion-time ordinates in (b) and (f) preserve extreme trials without compressing the median region.
This paper presents a learning-augmented trajectory planning framework for cooperative unmanned aerial vehicle (UAV) and unmanned ground vehicle (UGV) handover missions. While centralized trajectory optimization ensures dynamic feasibility and task optimality, its high computational cost limits real-time applicability. We propose a neural surrogate planner utilizing decoupled encoder-decoder long short-term memory (LSTM) networks to generate coordinated handover trajectory predictions from the task specifications. These predictions serve as informed warm starts for the downstream centralized optimizer, thereby accelerating convergence to dynamically feasible solutions. Benchmark evaluations demonstrate that the learning-augmented planning framework achieves more than a threefold speedup and 100% optimization success rate compared to cold start optimization. The results indicate that combining data-driven inference with model-based refinement enables fast and reliable trajectory generation for heterogeneous multi-robot systems.
Jingshan Chen, Bochen Yu, Henrik Ebel +1
Institute of Engineering and Computational Mechanics, University of Stuttgart, 70569 Stuttgart, Germany · Mechanical Engineering, LUT University, 53850 Lappeenranta, Finland
UAVs that act as relays or base stations need accurate air-to-ground path-loss predictions for rate adaptation and placement, but propagation conditions change as a UAV moves between suburban, urban and high-rise areas, and the same areas are often revisited. Online regressors that adapt by forgetting must relearn each environment from scratch, whereas a single model trained on all data averages incompatible regimes. We propose LiLib, a lightweight continual-learning scheme in which a UAV maintains a small library of recursive-least-squares experts. A windowed residual test detects drift; a short probe phase then either reuses the best stored expert or creates a new one. In simulations based on four standard urbanization profiles, LiLib reduces prediction RMSE from 5.89 dB (best sliding-window baseline) to 4.03 dB (p < 0.001), lowers the error shortly after a return to a known environment from 12.3 dB to 5.7 dB, and recovers 99% of the throughput of a regime-aware oracle in rate adaptation. The library stores four experts in under 0.5 KB, and identifies regimes with 92% purity without labels. When a second UAV is initialized with the library of a peer, its error after environment changes halves. LiLib does not reach the oracle, and similar regimes may be merged when shadowing is strong. The results indicate that, for recurring drift, remembering is more effective than re-adapting.
Minh Tran
Hanoi University of Science and Technology (HUST), Hanoi, Vietnam
Aerial image-goal navigation requires an unmanned aerial vehicle (UAV) to reach a target location specified by a goal image. Existing world-model-based methods rank candidate trajectories using predicted futures, but typically rely on only one or a few point predictions, which is inadequate for large-scale outdoor environments with substantial future-state uncertainty. To address this limitation, we propose the Uncertainty-Aware Navigation World Model (UA-NWM), an efficient latent world model for aerial image-goal navigation, which formulates trajectory scoring as conditional out-of-distribution detection. UA-NWM represents plausible futures with an uncertainty subspace and decomposes the prediction--goal discrepancy into uncertainty-explainable and unexplainable components. Only the unexplainable residual is used for scoring, enabling robust selection without multiple future samples. Extensive experiments demonstrate that UA-NWM consistently outperforms existing navigation world models while maintaining low inference latency. Real-world UAV experiments further validate its practical applicability. Project page: https://duryi.github.io/UA-NWM-Project-Page
Deyi Zhu, Haoyu Fan, Yinan Zhu +4
Tsinghua Shenzhen International Graduate School, Tsinghua University