Visualizing biomechanical exposure in augmented reality
The goal of this study is to design and develop an immersive approach by exploring augmented reality (AR) to deliver the knowledge regarding the biomechanical exposures to workers and to encourage workers to perform occupational tasks using appropriate body movements to reduce the risk of musculoskeletal disorders. This research is supported by NSF 1822477.
Inferring driving distraction through body movement
Driver distraction is a critical issue in transportation safety. In this study, we seek to apply deep neural networks on human kinematics data/images to infer driver’s distracted driving behaviors. The results can helpful for designing adequate distraction mitigation strategies in intelligent vehicles. This research is supported by NCSU Research and Innovation Seed Funding (RISF) [Video]
Mitigate the risks of musculoskeletal disorders during human-robot collaboration
In recent years, human-robot collaboration is becoming a thriving work configuration in which human workers and collaborative robots share the same workplace and work side by side. In this project, we seek to 1) use wearable sensors to infer worker’s full-body biomechanical exposure and 2) adjust the collaboration position between work and robot to reduce the exposure level. This research is supported by the National Institute for Occupational Safety and Health (T42-OH008673).
Vision-based ergonomic assessment through automated pose detection
The current assessment on MSD risk exposure predominantly relies on pen-paper based observational methods, which is very time-consuming and highly subjective to observers’ experiences. In the research, we seek to apply convolutional neural network (CNN) to automatically infer worker’s exposure level (i.e. RULA score) directly from images. This research is supported by the NC Occupational Safety & Health Education & Research Center (NC OSHERC) [Video].
Optimization-based biomechanical model for the shoulder joint
A biomechanical model can be helpful to understand the mechanical loading of musculoskeletal elements for injury prevention and clinical applications. In this study, an objective function including an entropy term is proposed to address muscle co-contractions. A musculoskeletal shoulder model is then developed to apply the proposed objective function for predicting shoulder muscle activities.
MOPED25: A multimodal occupational posture dataset
Kinematics data, as well as the synchronized videos of 12 subjects performing 25 different tasks, were collected with a motion tracking system. All the data was made publically available. Check out MOPED25 for further details.