Estimation of Ground Reaction Force Based on Computer Vision and Mobile Sensing for Floor Vibration Assessment

• Author / Creator

  Qian, Yuchen

• Floor vibrations caused by human activities, such as walking and running, should typically be
  addressed during the design phase. However, post-construction evaluations often fail to revisit
  these vibrations. This gap suggests a need for ongoing assessments to ensure that living
  experiences align with initial design intentions. To analyze the vibration levels of structures,
  accurately determining the Ground Reaction Force (GRF) is crucial, as it is one of the most critical
  components for identifying and predicting floor vibration serviceability. Estimating the floor
  response to vibrations requires precise input of the excitation load, which involves real-time data
  from human walking or running activities. In this study, the real-time GRF is estimated using
  mobile sensing and computer vision methods. The thesis is divided into three key parts. The first
  part involves developing a Computer Vision (CV) based method to estimate the human walking
  GRF on a treadmill. This method, based on OpenPose, a deep learning algorithm for multi-person
  pose estimation, detects key points of the human skeleton. The time-history of displacement from
  the middle waist is converted into GRF, and the results are compared with those obtained from
  professional wearable force measurement sensors (Loadsol). The findings indicate that the CV-
  based method can accurately measure real-time human walking GRF on a treadmill with a root
  mean square error (RMSE) of 10% in total.
  The second part of the thesis demonstrates the use of a smartphone accelerometer to measure GRF.
  Smartphone accelerometers can record the real-time acceleration of GRF, which can then be
  converted to real-time GRF. The output GRF is compared with that from the force measurement
  sensor Loadsol. The results show that the smartphone accelerometer method can accurately record
  the vertical GRF (with a RMSE of 9.6% in total) when Newton's second law is applied. In the final
  part, the smartphone mobile sensing method and CV estimation method are evaluated as a cost-
  effective and efficient alternative to the Loadsol reference. The thesis concludes with a discussion
  of the current work's limitations, recommendations, and potential directions for future research.

• Subjects / Keywords

  • Ground Reaction Force
  • Computer Vision
  • Construction
  • Design

• Graduation date

  Fall 2024

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-014a-r349

• License

  This thesis is made available by the University of Alberta Library with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.