Design and Implementation of Water Surface Cleaning Robot Based on Deep Learning
Most of the existing water surface cleaning devices cannot achieve autonomous cleaning for the entire water surface.This paper proposes a deep learning-based water surface cleaning robot.The robot is equipped with dual propeller motors,a link gathering garbage structure,and a crank push structure,which can complete autonomous cleaning and cruising of the water surface.Together with the vision system,it applies the convolutional neural network for training,and through the algorithms of backbone network,Ancher-Free detection head and loss function,it uses the YOLO PT model,and after 300 iterations,it realizes the garbage recognition accuracy of more than 87%in the self-constructed dataset.The robot also provides orientation data,which can achieve automatic mooring,providing a solution for the management of small urban water surfaces,promoting the utilization and recycling of garbage.
water surface cleaning robotmachine visiongarbage detectionYOLOautonomous mooring
万方数据
维普
