面向目标的语义通信在集成感知与通信的机器人避障中的应用

A goal-oriented semantic communication (GOSC) framework is proposed to enhance the integrated sensing and communication (ISAC) capabilities of unmanned aerial vehicles (UAVs) for obstacle avoidance tasks. This framework establishes a closed loop for sensing, command and control (C&C) generation, and transmission of sensing and C&C signals. For sensing, a Kalman filter (KF) is utilized to continuously predict the UAV's position, reducing the reliance on continuous sensing signal transmission and improving position estimation accuracy through sensing-prediction fusion. Based on the refined position estimation provided by the KF, a Mahalanobis distance algorithm is developed to further optimize obstacle avoidance decisions, thereby enhancing the UAV's autonomous navigation capabilities in complex environments. The effectiveness of this framework is validated through simulation experiments, demonstrating significant advantages in improving UAV obstacle avoidance performance.