Search

Computer vision has emerged as a promising tool for improving safety at construction sites through automatic scene recognition. However, traditional approaches require significant labor-intensive and time-consuming efforts for annotations. Although weakly supervised learning has the advantage of localizing objects without location information, the conventional class activation map (CAM) technique struggles with small and linear object localization and background noise at construction sites. For effective scene recognition related to construction safety, the spatial relationships between precisely localized objects are crucial. Due to the limitations of traditional CAM techniques in localizing small and linear objects, using CAM for construction safety monitoring remains inaccurate. Therefore, this study proposes a weakly supervised learning approach with an improved CAM for enhancing object detection in construction sites. The improved CAM localizes objects of various scales and is robust against background interference. Spatial relationships between localized objects are employed to determine the status of scenes for construction safety monitoring. Experiment using datasets associated with falls from ladders (FFL) demonstrates that the improved CAM surpasses the traditional CAM in mIoU (mean intersection over union) for the object localization performance as well as in the accuracy and F1-Score for recognizing unsafe scenes. This demonstrates the robust potential of employing CAM as a weakly supervised learning strategy, underlining its substantial feasibility for preventing hazards in construction sites. The proposed framework can minimize annotation efforts, demonstrating the potential of CAM as a viable computer vision technique for efficiently detecting hazards at construction sites.

In this study, we introduce a robust and affordable safety monitoring system using an object-detection algorithm trained using CutMix-based synthetic data. The image-processing algorithm, which is mainly used to improve productivity in the manufacturing industry, has recently emerged as a solution to safety problems at manufacturing sites. With deep learning, various applications have been developed to accurately monitor operator posture, wear safety equipment, and enter and exit hazardous areas. However, there are still limitations in terms of cost and time when using these applications in actual manufacturing sites, especially for small- and medium-sized enterprises. To address these challenges, our system offers the following optimal approaches: 1) leveraging domain-specific data-augmentation methods for efficient hand-monitoring systems, 2) implementing the entire system with accessible hardware and software, and 3) on-site validation at an actual small- and medium-sized enterprise, demonstrating superior performance over traditional augmentation methods. Our training method achieved 98.5% accuracy using just a minute of data, outperforming traditional methods by 28.7%. The system maintains an inference rate of over eight frames per second and can be configured for less than ${\$}200$ . Notably, our system allows rapid training and deployment in diverse environments and has a significant impact on manufacturing safety management.

This study presents a deep neural network (DNN)-based safety monitoring method. Nonstationary objects such as moving workers, heavy equipment, and pallets were detected, and their trajectories were tracked. Time-varying safety zones (SZs) of moving objects were estimated based on their trajectories, velocities, proceeding directions, and formations. SZ violations are defined by set operations with sets of points in the estimated SZs and the object trajectories. The proposed methods were tested using images acquired by CCTV cameras and virtual cameras in 3D simulations in plants and on loading docks. DNN-based detection and tracking provided accurate online estimation of time-varying SZs that were adequate for safety monitoring in the workplace. The set operation-based SZ violation definitions were flexible enough to monitor various violation scenarios that are currently monitored in workplaces. The proposed methods can be incorporated into existing site monitoring systems with single-view CCTV cameras at vantage points.

This paper presents a novel anchor rod force sensor based on fiber Bragg grating (FBG) for accurate anchoring force measurement. A special FBG strain sensor with enhanced sensitivity has been designed and embedded in the centre line of an elastic structural body, which forms the main body of the anchor rod force sensor. The mechanical structure design and measuring principle of the proposed anchor rod force sensor have been presented detailedly. Then a sensor prototype has been manufactured and fully tested. The experimental results show that the sensor provides a sensitivity of 11.58 pm/kN with a fitting linear correlation coefficient of 0.9996 in a measurement range of 50 kN, a repeatability error of 3.26%, and a hysteresis error of 3.87%. Furthermore, good resistance to bending effects of anchor rod, good creep resistance (fluctuant value is within 2 pm), and good temperature compensation capacity (maximum value is 11 pm) have also been verified. These results promise the sensor better applicability in the health monitoring of large civil engineering structures.

Task-oriented modular design and safety monitoring are demonstrated to be outstanding challenges for robots in practical application. In order to obtain better flexibility and better obstacles avoidance capability, design and analysis of a modular cable-driven parallel robot (MCDPR) are presented in this paper, which can be reconfigured to several configurations in engineering applications by changing the number of the mobile modules and the connection mode of the end effector. With regard to the motion control and safety monitoring system, the design and implementation of hardware and software are presented. An experimental prototype for the MCDPR is developed, followed by a brief illustration of the connection mode between the end effector and cables, and the installation of monitoring nodes. The performance of the MCDPR is discussed experimentally, including real-time pose monitoring and obstacle avoidance. The results verify the feasibility and efficiency of kinematics model and obstacle avoidance method, and the results also indicate that the designed motion control and safety monitoring system can realize trajectory tracking control, real-time safety monitoring, and obstacle avoidance for the MCDPR.

Low visibility hazard detection in construction sites is a crucial task for prevention of fatal accidents. Manual monitoring of construction workers to ensure they follow the safety rules (e.g., wear high-visibility vests) is a cumbersome task and practically infeasible in many applications. Therefore, an automated monitoring system is of both fundamental and practical interest. This paper proposes an intelligent solution that uses live camera images to detect workers who breach safety rules by not wearing high-visibility vests. The proposed solution is formulated in the form of an anomaly detection algorithm developed in the random finite set (RFS) framework. The proposed system is comprised of three steps: 1) applying a deep neural network to extract people in the image; 2) extracting particularly engineered features from each blob returned by the deep neural network; and 3) applying the RFS-based anomaly detection algorithm to each set of detected features. The experimental results demonstrate that in terms of F1-score, the proposed solution (as the combination of the newly engineered features and RFS-based anomaly detection algorithm) significantly outperforms various combinations of common and the state-of-the-art features and anomaly detection algorithms employed in machine vision applications.

When disturbances occur during high-speed train operations, real-time conflict prediction and operation re-planning play a critical role in guaranteeing the safety and efficiency of networked railway traffic. This paper attempts to propose a methodology to synergistically deal with the three critical aspects, i.e., safety monitoring, conflict resolution, and energy-efficient planning under the mechanism of model predictive control (MPC). A non-analytical rule-based high-level train movement model is employed to predict train movements and find out potential operation conflicts. It describes dynamic speed feedback adjustments under the precondition of keeping appropriate safety distances between trains. Conflict detection is implemented considering variation ranges of accelerations and decelerations to keep appropriate robustness. A predictive train rescheduling model is proposed incorporating the MPC mechanism and the non-analytical prediction model. After conflict prediction and resolution, the energy-efficient problems are formulated for overtaken trains toward junctions in a railway network. The corresponding solution algorithm is developed, taking into account the confinements of time-efficient movements of overtaking trains on energy-efficient movements of overtaken trains. Accordingly, operation synergies between overtaking and overtaken trains toward the junctions in a railway network can be achieved. An implementation algorithm is proposed, which holistically deals with robust safety monitoring and synergistic operation planning of high-speed trains in the prediction horizon, employing the train movement prediction model. Numerical experiments demonstrate the efficiency of the proposed methodology for train synergistic safe and efficient operations.