Figure - uploaded by Moein Younesi Heravi
Content may be subject to copyright.
Source publication
As the use of construction robots continues to increase, ensuring safety and productivity while working alongside human workers becomes crucial. To prevent collisions, robots must recognize human behavior in close proximity. However, single, or RGB-depth cameras have limitations, such as detection failure, sensor malfunction, occlusions, unconstrai...
Context in source publication
Context 1
... identify human activities associated with human-robot collaborative tasks in construction, this study examined existing construction robots, their functions, and the behaviors of collaborative human workers, as shown in Table 1. Although construction robots can perform several tasks with high accuracy, they still need the assistance of human co-workers to achieve a proper finish or ensure work quality. ...Similar publications
Understanding human behavior is a complex task that requires careful analysis. Researchers have been exploring various machine learning techniques to gain insights into human nature, but achieving high accuracy has been challenging. In our study, we propose using deep learning techniques to address this issue. To conduct our research, we collected...
Wearable Human Activity Recognition (HAR) is an important field of research in smart assistive technologies. Collecting the data needed to train reliable HAR classifiers is complex and expensive. As a way to mitigate data scarcity, Time Series Data Augmentation (TSDA) techniques have emerged as a promising approach for generating synthetic HAR data...
The technology for human activity recognition has diverse applications within the Internet of Things spectrum, including medical sensing, security measures, smart home systems, and more. Predominantly, human activity recognition methods have relied on contact sensors, and some research uses inertial sensors embedded in smartphones or other devices,...
Human Activity Recognition (HAR) using smartphone sensors has been identified as a significant emerging research domain. Its application areas exhibit the performance from the intelligent tailored activity monitoring. Researchers have proposed various HAR models to recognize the human activity patterns using traditional smartphone sensor data. In a...
Modern monostatic radar-based human activity recognition (HAR) systems perform very well as long as the direction of human activities is either towards or away from the radar. The monostatic single-input single-output (SISO) and monostatic multiple-input multiple-output (MIMO) radar systems cannot detect motion of an object that moves perpendicular...
Citations
... As a component of the data preprocessing for the ML models, the StandardScaler procedure was utilized. Its purpose is to normalize the functional scale of the data record 107,108 . This involves eliminating the average and clearing each parameter to www.nature.com/scientificreports/ ...
Graphene nanoplatelets (GrNs) emerge as promising conductive fillers to significantly enhance the electrical conductivity and strength of cementitious composites, contributing to the development of highly efficient composites and the advancement of non-destructive structural health monitoring techniques. However, the complexities involved in these nanoscale cementitious composites are markedly intricate. Conventional regression models encounter limitations in fully understanding these intricate compositions. Thus, the current study employed four machine learning (ML) methods such as decision tree (DT), categorical boosting machine (CatBoost), adaptive neuro-fuzzy inference system (ANFIS), and light gradient boosting machine (LightGBM) to establish strong prediction models for compressive strength (CS) of graphene nanoplatelets-based materials. An extensive dataset containing 172 data points was gathered from published literature for model development. The majority portion (70%) of the database was utilized for training the model while 30% was used for validating the model efficacy on unseen data. Different metrics were employed to assess the performance of the established ML models. In addition, SHapley Additve explanation (SHAP) for model interpretability. The DT, CatBoost, LightGBM, and ANFIS models exhibited excellent prediction efficacy with R-values of 0.8708, 0.9999, 0.9043, and 0.8662, respectively. While all the suggested models demonstrated acceptable accuracy in predicting compressive strength, the CatBoost model exhibited exceptional prediction efficiency. Furthermore, the SHAP analysis provided that the thickness of GrN plays a pivotal role in GrNCC, significantly influencing CS and consequently exhibiting the highest SHAP value of + 9.39. The diameter of GrN, curing age, and w/c ratio are also prominent features in estimating the strength of graphene nanoplatelets-based cementitious materials. This research underscores the efficacy of ML methods in accurately forecasting the characteristics of concrete reinforced with graphene nanoplatelets, providing a swift and economical substitute for laborious experimental procedures. It is suggested that to improve the generalization of the study, more inputs with increased datasets should be considered in future studies.
... Surveying, remote controlling of equipment and machinery, task performance, handling of materials (Jung et al, 2013, Jang et al, 2023, Sun et al, 2023, Hannan Qureshi et al, 2023 6. Wearable Technology Safety monitoring, worker health tracking, hazard or accident prevention, tools and equipment monitoring and productivity tracking. ...
SUMMARY: In recent years, the construction industry has witnessed a surge in the adoption and application of emerging technologies. Substantial evidence indicates the impact of these technologies on the lifecycle of construction projects. Among the notable categories of technologies embraced in construction, Information and Communication Technologies (ICTs) stand out. This category encompasses BIM models, reality capture, cloud computing, robotics, and more. However, a literature investigation that specifically focuses on the significant influence of these technologies on construction sites is lacking. To address this gap, this study conducted a systematic literature review, supported by a bibliometric analysis of published articles in the field of construction management. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) method, data was sourced from Scopus and Web of Science, covering the period from 2009 to 2023. The findings shed light on nine prominent ICT tools widely adopted on construction sites for project execution, with contributions from 113 countries emphasizing their profound impacts. Despite these positive aspects, this study also identifies notable challenges and provides recommendations for mitigating them. A smart site data integration framework was developed to facilitate the seamless integration of these technologies while ensuring data security. This study provides fundamental insights for future research in this field aimed at informing the construction industry about the importance of embracing cutting-edge technologies for enhanced efficiency on construction jobsites and for timely project delivery.
... As part of the pre-processing of the data for the ML models, the StandardScaler technique was employed. The role of the StandardScaler technique is to standardize the functional range of the database [174,175]. This is achieved by removing the mean and scaling each parameter to have a unit variance [176]. ...
Contemporary infrastructure requires structural elements with enhanced mechanical strength and durability. Integrating nanomaterials into concrete is a promising solution to improve concrete strength and durability. However, the intricacies of such nanoscale cementitious composites are highly complex. Traditional regression models encounter limitations in capturing these intricate compositions to provide accurate and reliable estimations. This study focuses on developing robust prediction models for the compressive strength (CS) of graphene nanoparticle-reinforced cementitious composites (GrNCC) through machine learning (ML) algorithms. Three ML models, bagging regressor (BR), decision tree (DT), and AdaBoost regressor (AR), were employed to predict CS based on a comprehensive dataset of 172 experimental values. Seven input parameters, including graphite nanoparticle (GrN) diameter, water-to-cement ratio (wc), GrN content (GC), ultrasonication (US), sand content (SC), curing age (CA), and GrN thickness (GT), were considered. The models were trained with 70 % of the data, and the remaining 30 % of the data was used for testing the models. Statistical metrics such as mean absolute error (MAE), root mean square error (RMSE) and correlation coefficient (R) were employed to assess the predictive accuracy of the models. The DT and AR models demonstrated exceptional accuracy, yielding high correlation coefficients of 0.983 and 0.979 for training, and 0.873 and 0.822 for testing, respectively. Shapley Additive exPlanation (SHAP) analysis highlighted the influential role of curing age and GrN thickness (GT), positively impacting CS, while an increased water-to-cement ratio (w/c) negatively affected CS. This study showcases the efficacy of ML techniques in accurately predicting CS of graphene nanoparticle-modified concrete, offering a swift and cost-effective approach for assessing nanomaterial impact on concrete strength and reducing reliance on time-consuming and expensive experiments.
... In the context of wearable sensor data, HAR is an important application area that involves associating data captured by sensors with specifi c activities or states. Depending on the data type, HAR can generally be divided into two categories: HAR for video data [7][8][9][10][11][12] and HAR for wearable sensor data [6,[13][14][15]. ...
... Camera-based action classifi cation methods use cameras or depth cameras to capture visual information about human movement. These methods typically involve computer vision and image processing techniques to enable real-time monitoring and classifi cation of human posture, and activities [10,12]. ...
The rapid development of wearable technology provides new opportunities for action data processing and classification techniques. Wearable sensors can monitor the physiological and motion signals of the human body in real-time, providing rich data sources for health monitoring, sports analysis, and human-computer interaction. This paper provides a comprehensive review of motion data processing and classification techniques based on wearable sensors, mainly including feature extraction techniques, classification techniques, and future development and challenges. First, this paper introduces the research background of wearable sensors, emphasizing their important applications in health monitoring, sports analysis, and human-computer interaction. Then, it elaborates on the work content of action data processing and classification techniques, including feature extraction, model construction, and activity recognition. In feature extraction techniques, this paper focuses on the content of shallow feature extraction and deep feature extraction; in classification techniques, it mainly studies traditional machine learning models and deep learning models. Finally, this paper points out the current challenges and prospects for future research directions. Through in-depth discussions of feature extraction techniques and classification techniques for sensor time series data in wearable technology, this paper helps promote the application and development of wearable technology in health monitoring, sports analysis, and human-computer interaction.
... It identifies the coordinates of the joints and employs a data fusion technique to derive the threedimensional locations of an object's body joints. To derive the 3D real-world coordinates, the module merges and transforms the 2D pixel joint locations detected frame-by-frame from each video using the particle filter algorithm, which was proposed by the authors in previous studies (Jang et al., 2023;Sarkar et al., 2022). The second module utilizes an LSTM-RNN model to classify the activity performed in each timeframe. ...
... We leveraged the power of multiple cameras to capture human actions from various angles, enabling robust 3D human pose estimation. The benefits and necessity of utilizing a multi-camera system in the context of 3D human pose estimation and activity recognition were extensively discussed in the authors' previous work (Jang et al., 2023). It is still worth mentioning here that while single-camera systems could be cost-effective and straightforward, they suffer from limitations such as the lack of depth information and susceptibility to occlusion. ...
... However, it is also important to acknowledge that the use of a multi-camera system comes with increased computational cost and financial investment compared to single-camera or RGB-D camera setups. Research has shown using four cameras could significantly increase data fusion time even compared to two cameras (Jang et al., 2023). The benefits of improved accuracy and robustness through more cameras must be carefully balanced with the associated computational load and financial implications. ...
Predicting human motion is a critical requirement in various applications, with particular significance in the construction sector. This task presents significant challenges due to the diverse nature of human actions and the complexities of converting 2D image coordinates to real-world space. In response to these challenges, this paper introduces an innovative deep learning-driven approach to forecast human motion trajectories, with a novel emphasis on activity recognition to improve predictive accuracy. The method utilizes a multi-camera system to extract 2D joint locations, which are then fused using a particle filter technique for 3D pose generation. Using 3D data, deep learning models are developed to first recognize the activity class, and then take it as auxiliary information for predicting the motion trajectory. Through a comprehensive experiment, we evaluated the proposed methodology. While the main innovation of the proposed approach lies in the incorporation of deep learning-based activity recognition into the trajectory prediction system, the experiment results revealed the activity-aware system’s capability to enhance prediction performance by a minimum of 6.4% and up to 16.6% in short-term forecasts, in compare with a conventional approach. Additionally, we analyzed the effects of varying time windows and joint selections on predictive outcomes across diverse scenarios and discussed the implications of these findings. By enhancing the prediction of human motion, this approach holds promise in improving workspace safety while encouraging effective interactions within complex environments.
The demand for steel fiber-reinforced concrete (SFRC) in construction has surged, particularly due to its enhanced fire resistance, leading to extensive research on its residual properties after elevated temperature exposure. However, conducting experimental tests can be a time-consuming process, demanding significant resources in terms of time, cost, and human resources. In contrast, machine learning (ML) models can rapidly simulate outcomes, enabling researchers to explore a wide range of scenarios efficiently. The previous literature studies used either ensemble or individual models; however, this study made a unique approach to utilize hybrid models, which are more precise compared to the other types of ML models. Accordingly, a data-rich framework containing 304 data samples was utilized in this study to develop an efficient and robust predictive model for the compressive strength (CS) of SFRC at higher temperatures. Support vector regression (SVR) in combination with three distinct optimization algorithms, specifically, particle swarm optimization (PSO), the firefly algorithm (FFA), and grey wolf optimization (GWO) were utilized to develop the hybrid models. In addition, typical ML techniques such as random forest (RF) and decision tree (DT) were used for comparative analysis. Among the five models, the SVR-FFA hybrid model showcased superior predictive accuracy, with the SVR-GWO model following closely. The correlation coefficients (R) for both models exceeded 0.99. The SVR-FFA model demonstrated relative root mean square error (RMSE) and mean absolute error (MAE) values of 0.0375 and 0.0248, respectively, while the SVR-GWO model exhibited corresponding values of 6.0560 and 5.1326. The SHapley Additive exPlanation (SHAP) technique revealed that the influence of temperature is more significant compared to the heating rate. Moreover, a considerable enhancement was observed in CS as the steel fiber volume fraction (Vf) reached 1.5%; on the other hand, no further improvement was noticed when Vf exceeded the 1.5% threshold. The proposed hybrid models are robust and accurate methods to estimate the CS of SFRC in field applications.
