•  
  •  
 

Journal of System Simulation

Abstract

Abstract: When a communication network is partially disabled or disrupted, an UAV is plunged into a "low-resource environment" and must rely on local hardware resources. This situation imposes constraints on computing power, storage capacity, and energy availability. To address the need for command intent recognition in such environments, a semi-physical simulation system for UAV in emergency rescue operations has been designed and implemented. Based on the low resource airborne hardware in the loop, the system simulates UAV command intention recognition and mission planning through GIS+BIM 3D environment modeling task scenarios. A new lightweight algorithm for intent recognition has been proposed, based on the joint attention fusion mechanism of global sentence structure information extraction using GraphSAGE and local semantic features of FastText, which optimizes and improves the accuracy and response speed of intent understanding prediction. On the constructed professional UAV command intent dataset, semi-physical simulation verifies that the accuracy of command intention recognition is 0.890 7 and the time is 58.808 ms, which meets the realtime requirement.

First Page

2894

Last Page

2905

CLC

TP391

Recommended Citation

Liu Hongfu, Fu Yajing, Zhang Wanpeng, et al. Algorithm and Semi-physical System Simulation for Command Intent Recognition of UAV in Low-resource Environment[J]. Journal of System Simulation, 2024, 36(12): 2894- 2905.

DOI

10.16182/j.issn1004731x.joss.24-0566

Share

COinS