This paper proposes an online signed distance function (SDF) learning-based safety-critical control method for an unmanned aerial vehicle (UAV). By utilizing LiDAR sensing data, the proposed method constructs an online SDF model of the environment, which transforms spatial geometric information into continuous distance constraints and further generates safety constraints. Subsequently, based on the high-order control barrier function (HOCBF) theory, the constructed safety constraints are incorporated into the UAV dynamics to achieve safe control in unknown environments. The proposed method provides geometrically meaningful distance information of the surrounding environment, effectively characterizes the spatial relationship between the UAV and obstacles, and is applicable to UAV system with complex high-order dynamic. Finally, simulation results demonstrate the effectiveness of the proposed method in guaranteeing safe UAV flight.

signed distance function (SDF),unmanned aerial vehicle (UAV),high-order control barrier function (HOCBF),unknown environment,LiDAR