Research on the Application of Multi-Sensor Net-works

Abstract

Single sensor sensing has insufficient coverage,poor anti-interference and limited detection accuracy,which is difficult to ensure the stable operation of intelligent sensing.This paper explores the multi-sensor fusion technology,and respectively builds the Carla vehicle obstacle avoidance system and the improved yolov5 strawberry disease recognition platform.The on-board system integrates millimeter wave radar,vision,differential positioning and can bus,and is developed based on Linux and python.It can avoid fixed obstacles, cross targets,front vehicle sudden braking and opposite vehicles,adapt to various road conditions in urban areas,high-speed and suburban areas,and help the automatic driving landing.Aiming at the pain points of low efficiency of strawberry artificial disease screening,frequent misdiagnosis and blind drug use,the system embedded CBAM attention to optimize small lesion extraction in yolov5,combined with manipulator and Jetson nano edge hardware,built agricultural equipment integrating disease identification,result display,prevention and control guidance, scientifically controlled pesticide consumption,and stabilized fruit output.The two systems rely on multi-source information complementation to make up for the short board of a single sensor,and customize the fusion scheme according to the application scenario,which is more cost-effective than general products.The experimental verification scheme is feasible,which can provide engineering reference for multi-sensor fusion in the field of intelligent transportation and modern agricultural detection.