Conference PaperPDF Available

Object Detection in Picking: Handling variety of a warehouse's articles

Authors:

Abstract

Purpose: The automation of picking is still a challenge as a high amount of flexibility is needed to handle different articles according to their requirements. Enabling robot picking in a dynamic warehouse environment consequently requires a sophisticated object detection system capable of handling a multitude of different articles. Methodology: Testing the applicability of object detection approaches for logistics research started with few objects producing promising results. In the context of warehouse environments, the applicability of such approaches to thousands of different articles is still doubted. Using different approaches in parallel may enable handling a plethora of different articles as well as the maintenance of object detection approach in case of changes to articles or assortments occur. Findings: Existing object detection algorithms are reliable if configured correctly. However, research in this field mostly focuses on a limited set of objects that need to be distinguished showing the functionality of the algorithm. Applying such algorithms in the context of logistics offers great potential, but also poses additional challenges. A huge variety of articles must be distinguished during picking, increasing complexity of the system with each article. A combination of different Convolutional Neural Networks may solve the problem. Originality: The suitability of existing object detection algorithms originates from research on automation of established processes in existing warehouses. A process model was already introduced enabling the transformation of laboratory trained CNNs to industrial warehouses. Experiments with CNNs according to this approach are published now.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
The first automated, high-bay, warehouses were introduced some 50 years ago. Since then, developments have continued at a rapid pace. Initially, automation was mainly focused on pallet warehouses with bulk storage facilities. A major reason was to increase the storage density, which could be achieved by making the warehouses higher. Later, mini-load warehouses and order picking warehouses were also automated. In this paper we will discuss the different types of automated systems as well as a number of scientific results that are now known about such systems. We will first discuss storage systems for unit loads (bins and pallets). This will be followed by order picking systems from which individual packages can be picked. Finally, we will provide our future expectations of warehouse automation.
Article
Full-text available
As the need for efficient warehouse logistics has increased in manufacturing systems, the use of automated guided vehicles (AGVs) has also increased to reduce travel time. The AGVs are controlled by a system using laser sensors or floor-embedded wires to transport pallets and their loads. Because such control systems have only predefined palletizing strategies, AGVs may fail to engage incorrectly positioned pallets. In this study, we consider a vision sensor-based method to address this shortcoming by recognizing a pallet’s position. We propose a multi-task deep learning architecture that simultaneously predicts distances and rotation based on images obtained from a visionary sensor. These predictions complement each other in learning, allowing a multi-task model to learn and execute tasks impossible with single-task models. The proposed model can accurately predict the rotation and displacement of the pallets to derive information necessary for the control system. This information can be used to optimize a palletizing strategy. The superiority of the proposed model was verified by an experiment on images of stored pallets that were collected from a visionary sensor attached to an AGV.
Article
Full-text available
Object detection and tracking is one of the most important and challenging branches in computer vision, and have been widely applied in various fields, such as health-care monitoring, autonomous driving, anomaly detection, and so on. With the rapid development of deep learning (DL) networks and GPU’s computing power, the performance of object detectors and trackers has been greatly improved. To understand the main development status of object detection and tracking pipeline thoroughly, in this survey, we have critically analyzed the existing DL network-based methods of object detection and tracking and described various benchmark datasets. This includes the recent development in granulated DL models. Primarily, we have provided a comprehensive overview of a variety of both generic object detection and specific object detection models. We have enlisted various comparative results for obtaining the best detector, tracker, and their combination. Moreover, we have listed the traditional and new applications of object detection and tracking showing its developmental trends. Finally, challenging issues, including the relevance of granular computing, in the said domain are elaborated as a future scope of research, together with some concerns. An extensive bibliography is also provided.
Conference Paper
Full-text available
Purpose: A picking system is presented ensuring order fulfilment and enabling trans-formation from manual to automated picking using a continuous learning process. It is based on Machine Learning for object detection and realized by a human-robot collaboration to meet requirements for flexibility and adaptability. A demonstrator is implemented to show cooperation and to evaluate the learning process. Methodology: The collaborative process, system architecture, and an approach for evaluation and workload balancing for order fulfilment and learning of robots during picking have already been introduced. However, a practical application is still miss-ing. A demonstrator is implemented using an agent-based architecture (JADEX) and a physical robot (UR5e) with a camera for object detection and first empirical data are evaluated. Findings: Single components of the demonstrator are already developed, but a pending task is to implement their interaction to analyze overall system perfor-mance. This work focuses on human-robot-interaction (Emergency Call), automated generation of images extended by feedback information, and training of algorithms for object detection. Requirements of human-machine interface, technical evalua-tion of image recording, and effort of algorithm training are discussed. Originality: Many approaches for automated picking assume a static range of ob-jects. However, this approach considers a changing range as well as a concept for transformation of manual to automated picking enabled by human-robot coopera-tion and automated image recording while enabling reliable order fulfilment.
Conference Paper
Full-text available
This work explores and compares the plethora of metrics for the performance evaluation of object-detection algorithms. Average precision (AP), for instance, is a popular metric for evaluating the accuracy of object detectors by estimating the area under the curve (AUC) of the precision × recall relationship. Depending on the point interpolation used in the plot, two different AP variants can be defined and, therefore, different results are generated. AP has six additional variants increasing the possibilities of benchmarking. The lack of consensus in different works and AP implementations is a problem faced by the academic and scientific communities. Metric implementations written in different computational languages and platforms are usually distributed with corresponding datasets sharing a given bounding-box description. Such projects indeed help the community with evaluation tools, but demand extra work to be adapted for other datasets and bounding-box formats. This work reviews the most used metrics for object detection detaching their differences, applications, and main concepts. It also proposes a standard implementation that can be used as a benchmark among different datasets with minimum adaptation on the annotation files.
Article
This study investigates the potentials and challenges of applying augmented reality (AR) smart glasses in logistics and supply chain management (SCM). A systematic literature review on AR smart glasses was conducted to provide a comprehensive synthesis of what has been published in the literature to capture the dynamics surrounding this technology and identify areas deserving of further academic attention. To summarise the latest developments in this field, eighty-two (82) publications were selected and thoroughly analysed. In terms of research findings, four main clusters of potential benefits were identified: visualisation, interaction, user convenience, and navigation. In contrast, the challenges posed by smart glasses in the logistics field were clustered around technical, organisational, and ergonomic considerations. The accumulation of knowledge and actionable insights in this study will benefit both academics and practitioners interested in this emerging wearable technology segment. As the first attempt to explore the importance of smart glasses in logistics and SCM activities, this study offers significant contributions to the literature by codifying extant knowledge on AR smart glasses and setting forth an agenda for future research.
Chapter
Convolutional neural network (CNN) has turned to be the state of the art for object detection task of computer vision. In this chapter, we have reviewed some popular state-of-the-art object detection models based on CNN. We have made a categorization of those detection models according to two different approaches: two-stage approach and one-stage approach. Herein, we have explored different gradual developments in two-stage object detection models from R-CNN to latest mask R-CNN as well as in one-stage detectors from YOLO to RefineDet. Along with the architectural description of different models, we have focused on training details of each model. We have also drawn a comparison among those models.