Determination of Automated Construction Operations from Sensor Data Using Machine Learning (original) (raw)

A Robust Framework for Identifying Automated Construction Operations

37th International Symposium on Automation and Robotics in Construction, 2020

Machine learning techniques have been successfully implemented for the identification of various construction activities using sensor data. However, there are very few studies on activity recognition in the automated construction of low-rise residential buildings. Automated construction is faster than conventional construction, with minimal human involvement. This requires high accuracy of identification for monitoring its operations. This paper discusses the development and testing of machine learning classifiers to identify normal automated construction operations with high precision. The framework developed in this work involves decomposing the activity recognition problem into a hierarchy of learning tasks in which activities at the lower levels have more details. The top recognition level divides the equipment states into two classes: 'Idle' and 'Operations'. The second recognition level divides the 'operations' into major classes depending on the top-level activities performed by the equipment. The third recognition level further divides the activities into subclasses and so on. Since the number of classes and the similarity between them increase with the recognition level, identification becomes extremely difficult. The identification framework developed in this study classifies operations belonging to the parent class at each level in the hierarchy. The efficacy of this framework is demonstrated with a case study of a top-down modular construction system. In this construction system, the modules of a structural frame are assembled and lifted starting with the top floor followed by the ones below. The accelerometer data collected during top-down construction is used to identify the construction operations. The proposed framework shows superior performance over conventional identification using a flat list of classes.

Inferring Construction Activities from Structural Responses Using Support Vector Machines

35 thInternational Symposium on Automation and Robotics in Construction, 2018

On-site data collection during construction activities help in evaluating productivity rates and preparing more accurate schedules. One of the challenges here is in collecting data automatically such that activity start times and durations can be computed reliably. This paper proposes a methodology to infer construction activities that are being performed on site using the structural responses collected from construction equipments. This methodology is applied to the case of a launching girder, an equipment used in the construction of viaducts in metro rail projects. There are four stages involved in the construction of a viaduct; Auto launching, Segment lifting, Post tensioning and Span lowering. Strain values from the launching girder are used to predict the stages of construction using machine learning techniques. Support Vector Machines are used to classify the strain data into one of the four classes corresponding the stage of construction. Data from a typical construction cycle is used for training. Using the model generated by the training data, subsequent activities can be inferred.

A hierarchical machine learning framework for the identification of automated construction

J. Inf. Technol. Constr., 2021

A robust monitoring system is essential for ensuring safety and reliability in automated construction. Activity recognition is one of the critical tasks in automated monitoring. Existing studies in this area have not fully exploited the potential for enhancing the performance of machine learning algorithms using domain knowledge, especially in problem formulation. This paper presents a hierarchical machine learning framework for improving the accuracy of identification of Automated Construction System (ACS) operations. The proposed identification framework arranges the operations to be identified in the form of a hierarchy and uses multiple classifiers that are organized hierarchically for separating the operation classes. It is tested on a laboratory prototype of an ACS, which follows a top-down construction method. The ACS consists of a set of lightweight and portable machinery designed to automate the construction of the structural frame of low-rise buildings . Accelerometers wer...

Prediction of Labor Activity Recognition in Construction with Machine Learning Algorithms

ICONTECH INTERNATIONAL JOURNAL, 2021

It is essential that the control and management of the work of labors in construction project management is effective. In this study, it is aimed to building artificial intelligence models to recognition on activities in a construction work to effectively utilization project management and control. In accordance with this purpose, 3-axis accelerometer, gyroscope, and magnetometer data were obtained from the labors through the sensor to predict the activities determined for a construction work. These raw data were made compliance for the model by going through a series of preprocessing applications. These data are trained and modeled with basic machine learning algorithms logistic regression, SVC, DT and KNN algorithms. According to the results of the analysis, the best prediction was obtained with the SVC algorithm with an accuracy of 90%. In other algorithms, respectively, 87% accuracy was contrived in the KNN algorithm, and approximately 80% accuracy in the logistic regression and...

Development of automated top-down construction system for low-rise building structures

International journal of industrialized construction, 2020

Automation is the best solution for achieving high productivity and quality in the construction industry at reduced cost and time. The main objective of this study is to develop an economical automated construction system (ACS) for low-rise buildings. The incremental development of the construction system and the structural system through different versions of laboratory prototypes are described in this paper. These ACS prototypes adopt a top-down construction method. This method involves the building of the structural system step by step from the top floor to the bottom floor by connecting and lifting structural modules. ACS prototype 1 consist of wooden structural modules and electric motor system. ACS prototype 2 has a highly automated custom designed hydraulic motor system to construct steel structural frame. ACS prototype 3 is a partially automated system where the steel structural modules are connected manually. These prototypes were evaluated on the basis of function, cost and efficiency of operations. Based on overall performance, ACS prototype 3 is identified as the best economical option for the construction of low-rise buildings. When the speed of construction is more important than cost, the ACS prototype 2 is the apt solution. This paper describes the challenges in developing an ACS and the criteria to evaluate its performance. It also includes a preliminary framework for the development of an automated construction monitoring system and its experimental evaluation. This machine learning-based framework is to identify the operations of ACS from sensor measurements using Support Vector Machines. Most of the operations are identified reasonably well and the best identification accuracy is 96%. The future studies are focusing on to improve the accuracy of operation identification, further development of the monitoring system and the ACS for actual implementation in construction sites.

Identification of the Structural State in Automated Modular Construction

36th International Symposium on Automation and Robotics in Construction, 2019

Automated construction involves complex interactions between machines and humans. Unless all possible scenarios involving construction and equipment are carefully evaluated, it may lead to failure of the structure or may cause severe accidents. Hence monitoring of automated construction is very important and sensors should be deployed for obtaining information about the actual state of the structure and the equipment. However, interpreting data from sensors is a great challenge. In this research, a methodology has been developed for monitoring in automated construction. The overall methodology involves a combination of traditional model-based system identification and machine learning techniques. The scope of this paper is limited to the machine learning module of the methodology. The efficacy of this approach is tested and evaluated using experiments involving the construction of a steel structural frame with one storey and one bay. The construction is carried out by a top-to-bottom method. During the construction of the frame, 99 base cases of normal operations are involved. 158 base cases of possible failures have been enumerated. Failure cases involve, for example, certain lifting platforms moving faster than others, improper connections of joints, etc. Strain gauges and accelerometers are installed on the structure and the data from these sensors are used to determine possible failure scenarios. Preliminary results indicate that machine learning has good potential for identifying activities and states in automated construction.

A Study of Supervised Machine Learning Techniques for Structural Health Monitoring

2015

We report on work that is part of the development of an agentbased structural health monitoring system. The data used are acoustic emission signals, and we classify these signals according to source mechanisms. The agents are proxies for communicationand computation-intensive techniques and respond to the situation at hand by determining an appropriate constellation of techniques. It is critical that the system have a repertoire of classifiers with different characteristics so that a combination appropriate for the situation at hand can generally be found. We use unsupervised learning for identifying the existence and location of damage but supervised learning for identifying the type and severity of damage. This paper reports on results for supervised learning techniques: support vector machines (SVMs), naive Bayes classifiers (NBs), feedforward neural networks (FNNs), and two kinds of ensemble learning, random forests and AdaBoost. We found the SVMs to be the most precise and the ...

APPUCATIONS OF MACHINE LEARNING TO CONSTRUCTION SAFElY

1993

This paper discusses potential applications of machine learning in construction safety. Both learning about accidents and their prevention are described, including examples which demonstrate practical applications of machine learning. Examples were developed using actual aCCident records and two learning systems: ROUGH. based on the theory of rough sets. and IN LEN . based on the STAR methodology. The paper also discusses future research needs and directions.