H. Junker, P. Lukowicz, and G. Tröster (Switzerland)
Activity Recognition, Wearable Computing, Locomotion Analysis
The paper describes a novel method for the classification of modes of locomotion that can be implemented using particularly low power wearable sensors, and low com plexity feature extraction and classification algorithms. It is based on the analysis of the time lag between initial heel and ball contact of the foot. The lag is computed from two force sensitive resistors, appropriately mounted under the sole. The paper discusses the biomechanical considerations behind our method and presents an exper imental evaluation using the example of level walking and stair climbing/descending. The results suggest that the de scribed method is efficient in classifying different modes of locomotion on a step basis.
Important Links:
Go Back
