Human Activity Recognition (HAR)

HAR is about the understanding of human activities and their surrounding environment via human-generated signals (IMU Sensor).

We currently have the following research topics within the area

  1. Utilizing IMU to collect vibration during wheelchair usage to classify road conditions, which can then be helpful in travel planning (i.e. locating a more wheelchair-friendly road) for wheelchair users. For more detail about this sub-topic, please check my other project [MyPath].
  2. Detection of incorrect glasses usage for elderly glasses users to avoid potentially fatal conditions like falling down arising from using wrong glasses.
  3. Using UWB sensors to improve the recognition accuracy of small-movement activities without camera usage: for instance, pulling a book from a shelf can be considered a small-movement activity, and improving accuracy is about correctly identifying which book was pulled.