Agentic AI-Enabled Robust Sensing for ISAC Low-altitude MEC Networks: Challenges, Solutions, and Case Study

The integration of low-altitude mobile edge computing (MEC) and integrated sensing and communication (ISAC) holds immense potential for empowering applications such as aerial inspection and emergency rescue. However, compared to terrestrial networks, low-altitude environments pose severe challenges to sensing robustness due to high dynamics, complex channel clutter, and resource constraints. In this article, we investigate Agentic AI-based solutions to achieve robust sensing in low-altitude ISAC-MEC networks (ISAC-MECs). Specifically, we first analyze the core challenges of achieving sensing robustness in low-altitude scenarios and discuss the limitations of conventional AI methods in addressing these complexities. Then, we introduce the modular architecture and workflow of Agentic AI, followed by an analysis of its role in addressing sensing robustness challenges. To demonstrate the practical benefits of Agentic AI for robust sensing in low-altitude ISAC-MECs, we propose a physics-reasoning-enhanced Agentic AI-based optimization framework via a representative case study. The framework uses potential-guided and self-refinement mechanisms to iteratively refine state representations and reward functions, which improves reinforcement learning performance for robust sensing in low-altitude ISAC-MECs. Simulation results validate the effectiveness of the proposed framework. Finally, we outline future directions for integrating Agentic