Integrated Sensing, Communication and Control with Deep Reinforcement Learning for Low-Altitude Intelligent Transportation

With the accelerating pace of global urbanization, the development of Low-Altitude Intelligent Transportation (LAIT) systems powered by electric Vertical Take-Off and Landing (eVTOL) aircraft has become a critical strategic direction to mitigate urban development pressures. However, the operational complexity of eVTOLs in dense urban environments, combined with their high-speed maneuverability, presents significant tech-nical challenges for LA IT systems, including frequent communication link interruptions and difficulties in achieving coordinated control among multiple e VTOLs. To address these challenges, this paper proposes a novel deep reinforcement learning-based communication-sensing cooperative control approach that seam-lessly integrates beam prediction, base station handover, and trajectory planning into a unified approach. The proposed approach employs adaptive beamforming and intelligent handover mechanisms to dynamically optimize communication quality in real time, while incorporating communication performance as a key constraint within the trajectory planning process. This enables a tightly coupled optimization of communication resources and flight trajectories, overcoming the limitations of conventional decoupled communication and control designs. The result is a closed-loop system that simultaneously enhances communication reliability and flight performance. Simulation results demonstrate that the proposed approach significantly enhances communication quality during eVTOL operations while maintaining reliable connectivity and improving transportation efficiency.