Abstract:The advancement of automation and Internet of Things technology has bolstered the automation process in the logistics sector. To address the challenge of localizing and generating grasping positions for intelligent robots in logistics sorting, this study developed an algorithm for item localization. The algorithm relies on enhanced YOLOv3 target detection and instance segmentation technologies to design a position generation algorithm for the robotic arm, which was further refined using sampling evaluation. The experimental results showed that the research-improved target detection model performed better on different datasets in terms of F1 value, accuracy and Area under the Curve (AUC) metrics, with the highest values of 95.77%, 94.05%, and 91.30%, respectively, which was effective in localizing document-like parcels. Meanwhile, the instance segmentation algorithm with fused features took significantly lower values than other target detection algorithms in terms of average absolute value error and root mean square error. The accuracy rate and all-class average precision value were higher than other target detection models, and the fluctuation of the value taken was smaller, which was suitable for logistics parcel localization. The position generation model, based on a sampling evaluation, yielded significantly different values compared to other algorithms. The relative position error and absolute trajectory error indexes were all below 0.4. The combined indexes of grasping accuracy and error indicate the superior performance of the research-designed algorithms. They can effectively enhance the sorting effects of real logistics scenarios. This research contributes to the improvement of the automated sorting system through the use of visual robotic arm technology. Additionally, it encourages the development of logistics automation and the establishment of intelligent logistics factories.

Hardware Oriented Vision System of Logistics Robotics

A Smart Surface Inspection System Using Faster R-CNN in Cloud-Edge Computing Environment

A CNN-Based Visual Sorting System with Cloud-Edge Computing for Flexible Manufacturing Systems

Rapid Detection of Moving Target in the Competitive Networked Robots

Game Robot'S Vision Based On Faster Rcnn

Fast Visual Identification and Location Algorithm for Industrial Sorting Robots Based on Deep Learning

An Efficient Real-Time Object Detection Framework on Resource-Constricted Hardware Devices via Software and Hardware Co-design

Fasteners quantitative detection and lightweight deployment based on improved YOLOv8

Research on the Algorithm of Position Correction for High-Speed Moving Express Packages Based on Traditional Vision and AI Vision

Deep Learning for Precise Robot Position Prediction in Logistics

CNN and Binocular Vision-Based Target Detection and Ranging Framework of Intelligent Railway System

Efficient Visual Fault Detection for Freight Train via Neural Architecture Search with Data Volume Robustness

Lightweight convolutional neural network for fast visual perception of storage location status in stereo warehouse

Vision-based robotic grasping using faster R-CNN–GRCNN dual-layer detection mechanism

Deep-Learning-Based Accurate Identification of Warehouse Goods for Robot Picking Operations

Research on high-speed classification and location algorithm for logistics parcels based on a monocular camera

Visual Sorting of Express Parcels Based on Multi-Task Deep Learning

Real-time Cargo Loading Identification Based on Self-powered Camera

An Improved Lightweight Deep Learning Model and Implementation for Track Fastener Defect Detection with Unmanned Aerial Vehicles

Design of Logistics Sorting Algorithm Based on Deep Learning and Sampling Evaluation

A Design of Smart Unmanned Vending Machine for New Retail Based on Binocular Camera and Machine Vision