A Reinforcement Learning Approach to View Planning for Automated Inspection Tasks

Manual inspection of workpieces in highly flexible production facilities with small lot sizes is costly and less reliable compared to automated inspection systems. Reinforcement Learning (RL) offers promising, intelligent solutions for robotic inspection and manufacturing tasks. This paper presents an RL-based approach to determine a high-quality set of sensor view poses for arbitrary workpieces based on their 3D computer-aided design (CAD). The framework extends available open-source libraries and provides an interface to the Robot Operating System (ROS) for deploying any supported robot and sensor. The integration into commonly used OpenAI Gym and Baselines leads to an expandable and comparable benchmark for RL algorithms. We give a comprehensive overview of related work in the field of view planning and RL. A comparison of different RL algorithms provides a proof of concept for the framework’s functionality in experimental scenarios. The obtained results exhibit a coverage ratio of up to 0.8 illustrating its potential impact and expandability. The project will be made publicly available along with this article.

Download Full-text

A Solution to the Missing-View Problem for Polyhedral Solids

17th Design Automation Conference: Volume 2 — Computer-Aided Design, Mechanical Systems Simulation, and Analysis, Mechanisms, and Robotics ◽

10.1115/detc1991-0126 ◽

1991 ◽

Author(s):

Y. L. Srinivas ◽

Debasish Dutta

Keyword(s):

Machine Vision ◽

Computer Aided Design ◽

Partial Information ◽

Connectivity Matrix ◽

Automated Inspection ◽

Vertex Connectivity ◽

Computer Aided ◽

Inspection Systems ◽

Aided Design ◽

The Given

Abstract An algorithm for generating the missing view corresponding to a given pair of orthoghonal views of a polyhedral solid is presented. The solution involves reconstructing the solids from the partial information given and then generating the missing view. The input is a vertex connectivity matrix describing the given views. Reconstruction of solids from incomplete orthographic views will have applications in computer-aided design, machine vision and automated inspection systems.

Download Full-text

Taking the Mechanical Pulse

Mechanical Engineering ◽

10.1115/1.2004-may-3 ◽

2004 ◽

Vol 126 (05) ◽

pp. 33-35 ◽

Cited By ~ 1

Author(s):

Jean Thilmany

Keyword(s):

Data Acquisition ◽

Computer Aided Design ◽

National Laboratory ◽

Argonne National Laboratory ◽

Data Acquisition Systems ◽

Vision Inspection ◽

Data Acquisition Software ◽

Inspection Systems ◽

Mechanical Pulse ◽

Aided Design

This article provides details of various applications of data acquisition systems. As data acquisition hardware is coupled with the software, which users can adapt for their own unique applications, data acquisition systems can be configured to fulfil a range of purposes. They are used for test and measurement and for industrial automation, and can serve as the eyes of a production line or the nose of a sensor. At Innoventor Inc., St. Louis, engineers have created vision inspection systems and pick-and-pack equipment for customers; they’ve designed machine control systems and robotics. According to an engineer in the company, data acquisition systems are a check on the confidence that today’s computer-aided design and analysis software engender. Data acquisition systems can be customized for a testing situation or environment. In addition to acquiring data from prototypes, a system can be configured to measure products on a manufacturing line or measure the line itself. Researchers at Argonne National Laboratory in Illinois used data acquisition software and hardware to develop their Smart Sensor Developer Kit, a chemical microsensor that can identify almost any air bound gaseous chemical.

Download Full-text

Complex Shape Measurement Using 3D Scanner

Jurnal Teknologi ◽

10.11113/jt.v45.333 ◽

2012 ◽

Cited By ~ 2

Author(s):

Rahmita Wirza O.K. Rahmat ◽

Seng Beng Ng ◽

Kamini Sangaralingam

Keyword(s):

Computer Aided Design ◽

Complex Shape ◽

Knee Prosthesis ◽

Three Dimensional ◽

Laser Scanner ◽

Raw Material ◽

Surface Model ◽

Automated Inspection ◽

3D Scanner ◽

Inspection Systems

Sistem Pemeriksaan Automatik biasanya bermula dengan teknik Reka bentuk Berbantukan Komputer (CAD) dan berakhir dengan janaan arahan mesin untuk menukarkan bahan mentah kepada produk akhir, memperolehi data tiga dimensi untuk membina model permukaan berparameter bagi tujuan analisis ralat, duplikasi atau memperbaiki objek tersebut. Ia melibatkan pendigitalan permukaan objek yang perlu diperiksa, janaan model CAD dan analisis keputusan pemeriksaan. Pendigitalan boleh dilakukan dengan teknik kuar sentuh atau pengesan tanpa sentuh. Biasanya, tanpa bantuan daripada pengeluar, adalah sukar untuk memantau dan mengenal pasti perubahan produk selepas digunakan. Tambahan pula, reka bentuk cetakan asal tidak didedahkan. Objektif utama kertas ini ialah mencadangkan suatu algoritma yang mampu mengekstrak maklumat daripada objek berbentuk kompleks yang telah didigitalkan (dengan menggunakan pengimbas laser), misalnya cangkukan lutut, untuk menjanakan reka bentuk cetakannya. Diharapkan penyelidik lain, terutamanya dalam bidang perubatan dan bio-kejuruteraan mendapat manfaat daripada algoritma ini untuk meramal dan menganggar perubahan bekas cangkukan perubatan dan seterusnya mengubah reka bentuknya untuk memenuhi kehendak pesakit. Kata kunci: Pemeriksaan automatik, reka bentuk cetakan, cangkukan lutut, pengimbas 3D, gantian lutut total Automated inspection systems usually start from computer-aided design (CAD) techniques, and end with either generation of machining instructions required to convert a raw material into a finished product, or to obtain three-dimensional data for constructing a parametric surface model of the product for the purpose of error analysis, or to duplicate or enhance the object. It involves surface digitization of an existing part that needs inspection, CAD model creation and analyzing the inspection results. Most of the time, without the help from the manufacturer, it is very difficult to do inspection and to detect any changes of the product after being used. In addition, the original blueprints of most manufacturing products are not being revealed to the public. The main objective of this paper is to propose an algorithm that is able to extract information from digitized complex shape objects (using laser scanner), such as the medical knee implant (knee prosthesis), to generate their blueprint. It is hoped that other researchers, especially in the medical and bioengineering field, can benefit from this proposed algorithm to predict or approximate the changes of the medical ex-plant and modify its design to suit the patient needs. Key words: Automated inspection, blueprint, prosthesis, 3D scanner, total knee replacement

Download Full-text

An Overview on the Latest Nature-Inspired and Metaheuristics-Based Image Registration Algorithms

Applied Sciences ◽

10.3390/app10061928 ◽

2020 ◽

Vol 10 (6) ◽

pp. 1928 ◽

Cited By ~ 1

Author(s):

J. Santamaría ◽

M. L. Rivero-Cejudo ◽

M. A. Martos-Fernández ◽

F. Roca

Keyword(s):

Image Registration ◽

Computer Aided Design ◽

Computational Models ◽

State Of The Art ◽

General Purpose ◽

Metaheuristic Algorithms ◽

Comprehensive Overview ◽

Multiple Viewpoints ◽

Automated Image Registration ◽

Aided Design

The development of automated image registration (IR) methods is a well-known issue within the computer vision (CV) field and it has been largely addressed from multiple viewpoints. IR has been applied to a high number of real-world scenarios ranging from remote sensing to medical imaging, artificial vision, and computer-aided design. In the last two decades, there has been an outstanding interest in the application of new optimization approaches for dealing with the main drawbacks present in the early IR methods, e.g., the Iterative Closest Point (ICP) algorithm. In particular, nature-inspired computation, e.g., evolutionary computation (EC), provides computational models that have their origin in evolution theories of nature. Moreover, other general purpose algorithms known as metaheuristics are also considered in this category of methods. Both nature-inspired and metaheuristic algorithms have been extensively adopted for tackling the IR problem, thus becoming a reliable alternative for optimization purposes. In this contribution, we aim to perform a comprehensive overview of the last decade (2009–2019) regarding the successful usage of this family of optimization approaches when facing the IR problem. Specifically, twenty-four methods (around 16 percent) of more than one hundred and fifty different contributions in the state-of-the-art have been selected. Several enhancements have been accordingly provided based on the promising outcomes shown by specific algorithmic designs. Finally, our research has shown that the field of nature-inspired and metaheuristic algorithms has increased its interest in the last decade to address the IR problem, and it has been highlighted that there is still room for improvement.

Download Full-text