Vibration-based Terrain Classification Recognition Using a Six-axis Accelerometer
Résumé
Aging is a key problem in our daily life which demands more technology such as the terrain classification recognition for the detailed path planning. In this article, a six-axis accelerometer was placed on the base of a walking stick to capture walking gait vibration information. The signal features in the collected information were used to reflect the terrain. We used a matching tracking method to decompose the signal data into linear combinations of Gabor atoms and a joint time-frequency analysis method to perform analysis on the topographic features in a three-dimensional space. Three types of terrains were classified from space-time data. Results show that the proposed algorithm could achieve an average accuracy rate of 88.6%, which is 4% higher than conventional methods.
Origine | Fichiers produits par l'(les) auteur(s) |
---|