Answers are grounded in this paper and cite their sources.
With the emergence of the low-altitude economy, unmanned aerial vehicles (UAVs) have become pivotal components in mobile edge computing (MEC) architectures, leveraging their mobility and line-of-sight (LoS) advantages for computation offloading. However, autonomous operation in complex environments requires addressing two critical challenges: generating collision-free trajectories amidst dense obstacles while ensuring fair resource allocation among heterogeneous users with time-varying demands. This paper presents a unified agentic artificial intelligence (AI) framework that jointly optimizes obstacle avoidance, trajectory planning, and fairness-aware scheduling. By treating the UAV as an autonomous agent, the framework endows the system with goal-driven autonomy through five collaborative modules, namely perception, decision, execution, self-evaluation, and memory, enabling continuous closed-loop operation rather than relying on manually designed rules. Deep reinforcement learning (DRL), specifically the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm, serves as the underlying engine. The state representation integrates spatial localization, obstacle proximity, and service history, while the reward function balances fairness, user energy consumption, and collision avoidance. Comparative evaluations against Deep Deterministic Policy Gradient (DDPG) and Soft Actor-Critic (SAC) demonstrate 5%–10% improvements in fairness while maintaining collision-free operation. Results validate the effectiveness of the proposed agentic AI framework for autonomous UAV operations in complex low-altitude environments.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: