All about Lab
Structural Control and Intelligent Systems Laboratory

SCaIS

• • The vision of the Structural Control and Intelligent Systems Laboratory (SCaIS Lab.) is to establish hazard-resilient communities by making civil infra-structures intelligent. To achieve the vision, the SCaIS Lab. is trying to develop innovative smart structures technology and apply it to civil infra-structures. Especially, we are focusing on smart vibration control systems based on smart materials such as MR fluid/elastomer and energy harvesting techniques such as electromagnetic induction transducer. Also, the studies on application of advanced sensing/actuation technologies such as wireless sensor networks and piezoelectric sensors/actuators to civil infra-structures have been recently started. To effectively perform their research activities, the SCaIS Lab. is collaborating with research teams within KAIST as well as those outside in various engineering fields, including mechanical engineering and electrical and electronics engineering.

Research Themes

• • SCaIS Lab advances resilient and intelligent civil infrastructure across four thrusts, ranging from three-dimensional seismic isolation with smart, adaptive damping materials to data-driven monitoring, autonomous inspection, and AI-enabled jobsite safety. We design 3D isolation systems and smart dampers to actively control tri-axial seismic response, validating performance through full-scale shaking-table testing for safer structures. In parallel, we develop an earthquake decision-support framework that couples offline reduced-order models with online sensor assimilation to deliver real-time condition assessment and continuously updated fragility curves. For inspection, we deploy cooperative multi-UAV master–slave architectures that achieve precise localization in GNSS-challenged settings and a hierarchical AI pipeline that detects and segments damage efficiently at scale. Finally, we translate AI to construction safety via real-time vision modules for PPE compliance, equipment tracking, and hazard prediction, strengthened by domain-adaptive training for robust performance at small and medium sites.
  • The major research topics at SCAIS include, but are not limited to:

• 1.

  Developing a Seismic isolation system using smart materials

• 2.

  Data-Driven Monitoring and Decision Support for Earthquake Response

• 3.

  Autonomous Infrastructure Monitoring Through Cooperative UAV Systems and Hierarchical AI Analysis

• 4.

  Smart Technologies for Improving Safety at Construction Sites

Research Topics

• 1) Developing a Seismic isolation system using smart materials
  • • Mitigating the damage to both structural and nonstructural elements during seismic events is a fundamental challenge in civil engineering. Conventional seismic isolation systems are mainly developed for horizontal base isolation and have limitations in the ability to effectively control three-axis seismic response. In order to overcome these limitations, our lab is developing and studying three-dimensional seismic isolation systems. In addition, advanced damping systems are being developed that utilize smart materials capable of adapting to diverse seismic inputs, thereby reducing seismic responses and enhancing structural safety. SCAIS LAB conducts research from smart material development to isolation system design, and evaluates performance through full-scale shaking table experiments, with the goal of safer structural systems.

    

    Seismic Isolation System: From Design and Modeling to Experimental Testing

• 2) Data-Driven Monitoring and Decision Support for Earthquake Response
  • • Modern critical infrastructure, such as nuclear power plants and large civil structures, requires faster and smarter ways to respond to earthquakes. Traditional approaches often rely on manual inspections, which can take a long time and delay recovery. Our research introduces a data-driven monitoring and decision support framework designed to provide immediate and reliable safety evaluations. The framework combines two phases. In the offline phase, detailed computer simulations are performed to build simplified but accurate models of structural behavior under different damage scenarios. These reduced-order models, or ROMs, allow for rapid calculations without losing essential accuracy. In the online phase, real earthquake data are collected directly from sensors installed on the structure. This information is continuously compared with the precomputed models to update the system’s condition in real time. As shown in the figures, this integration enables not only quick verification of structural safety but also updating of fragility curves—probabilistic tools that estimate the likelihood of different levels of damage given earthquake intensity. By shifting from a “reactive” approach, where problems are only identified after visible damage, to a “predict and prevent” strategy, this framework improves safety, minimizes downtime, and reduces unnecessary shutdowns. In practice, this means that after an earthquake, decision makers can quickly determine whether a facility is safe to continue operation or needs intervention, supported by data-driven evidence rather than lengthy manual checks.

    

    Earthquake-informed monitoring and decision-support framework (Chanwoo Lee, Ph. D thesis, 2025)

• 3) Autonomous Infrastructure Monitoring Through Cooperative UAV Systems and Hierarchical AI Analysis
  • • Inspecting large-scale infrastructure is hazardous, slow, and costly, and autonomous UAVs struggle with precise localization in GNSS-challenged environments; to overcome this, we employ a cooperative multi-UAV Master–Slave architecture in which a Master with stable satellite connectivity localizes and coordinates Slave UAVs via inter-vehicle links for accurate positioning and tasking. After autonomous data capture, we process high-resolution imagery with a hierarchical AI pipeline: a fast, lightweight detector sweeps entire images to flag likely damage, and a compute-intensive segmentation model then focuses only on those regions for detailed analysis and quantification. This approach preserves assessment accuracy while minimizing computation, delivering an end-to-end solution for intelligent monitoring and objective damage metrics that enable predictive maintenance.
    • 

      Cooperative Multi-UAV System for Bridge Inspection

    • 

      Hierarchical AI Model Architecture for Damage Detection

• 4) Smart Technologies for Improving Safety at Construction Sites
  • • The construction industry remains one of the most hazardous sectors, with accident rates consistently higher than other industries. This study presents an integrated framework of smart safety monitoring technologies to address these risks, particularly at small and medium sized sites where safety budgets and manpower are limited. Image-based AI modules, including hard-hat wearing detection, heavy equipment operation tracking, signal-worker arrangement recognition, collision risk detection, and hazardous area access monitoring, were developed using YOLOv5, DeepSORT, and ResNet-based models. Their field application demonstrated strong performance in real-time hazard detection under complex and far-field conditions. Furthermore, a domain adaptation strategy combining self-training, reconstruction, and adversarial learning was introduced to overcome site-to-site variations, with Cross-Hard Negative Mining (Cross-HNM) yielding significant improvements in generalization. The results indicate that AI-enabled monitoring systems, coupled with adaptive training strategies, can substantially enhance the practicality and effectiveness of safety management in construction environments.
    • 

      Overview of the training strategy for domain-adaptive object detection (Kim et al. 2024)

    • 

      Schematic of the real-time struck-by hazards detection system (Kim et al. 2023)

Recent News

• [Jul. 2025]
  • Dr. Seung-Kyung Kye, an alumnus of our lab, has been appointed to the position of Staff Researcher at the Korea Authority of Land & Infrastructure Safety.

• [Jun. 2025]
  • Dr. Hyung-Soo Kim, an alumnus of our lab, has been appointed to the position of Staff Researcher at the KEPCO Research Institute, KEPRI.

• [Jul, 2025]
  • Ph. D. student Jaehwan Seong was selected for a Plenary Talk at ISARC 2025(42nd International Symposium on Automation and Robotics in Construction).

• [Jul. 2025]
  • Prof. Hyung-Jo Jung gave an invited speech at ANCRiSST2025 & ZHITU2025 (The Joint 16th International Workshop on Advanced Smart Materials and Smart Structures Technology and 4th ZHITU Symposium on Advances in Civil Engineering)

• [Jun. 2024]
  • Dr. Jae-Chan Park, an alumnus of our lab, has been appointed to the position of Staff Researcher at the Korea Institute of Nuclear Safety.