The applied structural dynamics lab (ASDL) analyzes the dynamic response of civil infrastructure such as bridges, power plant structures, and buildings through the rigorous dynamic analysis and advanced real-time testing. The major research interests of ASDL include, but not limited to, structural vibration control, seismic performance evaluation of structures, nonlinear dynamic analysis, and real-time hybrid simulation (RTHS). In particular, RTHS is a very efficient testing method for analyzing the response of large-scale structures under earthquake loads. Since only the structural component of interest, which is difficult to be analytically analyzed, is physically tested and the rest of the structure is numerically modeled, RTHS enables a large-scale dynamic testing with a relatively low cost. The substructuring concept of RTHS can be extended to the areas of wind engineering and coastal engineering so that the aeroelastic behavior of bridges or the coastal structure subjected to both wind and wave loads can be also effectively analyzed using RTHS. Furthermore, ASDL will expedite the development of robots that can be used in a high-risk construction site or disaster area by using its unique real-time displacement/force control algorithms which are the core elements for high-speed motion control of humanoid robots.
<Example of RTHS for a 2-span bridge>
Multi-axis RTHS for floating bridges subjected to both wind and wave loads
Improvement of real-time displacement and force control algorithms
Development of a framework for 3-dimensional large-scale RTHS
Multi-axis RTHS for isolated buildings with friction pendulum system (FPS)
Deep learning for seismic response prediction and damage assessment of nuclear power plant structures and mechanical equipment
Development of new system identification methods using multi-tier sensor network for nuclear power plant structures
Experimental validation of deep learning technologies by conducting large-scale hybrid simulations
Model Predictive Control / Whole-Body Control
Reinforcement Learning (RL) for analyzing gaits, running patterns, and their smooth transitions
Real-time force control for highly dynamic motions
Development of high-power torque motors integrated with real-time force & displacement control modules
SLAM (Simultaneous Localization and Mapping)
Autonomous motion control using LiDAR and optical camera
Bipedal and quadrupedal robots have been developed brilliantly so far, but these robots still have difficulty to realize the highly dynamic motions of human and animals
Each leg of cheetah only touches the ground for about 30~40msec in a single cycle of sprint motion, where the impulsive loading is accurately controlled for chasing the prey. New technologies for real-time force/displacement control will enable highly dynamic bipedal and quadrupedal robots that can even run as fast as cheetah.
Large-scale 3-dimensional earthquake loading application system integrated with real-time multi-axis displacement and force controls
Maximum axial loading capacity > 15,000kN
Maximum lateral loading capacity > 3,000kN