Three-Dimensional Direction Recognition Using FSR Sensors

Attaching a serial of small tactile sensors to an existing robotic system is relatively easy compared to integrating most other tactile arrays designs. This paper proposes a method to recognize human intended direction in three-dimensional space using a spherical handlebar equipped with tactile sensors. Some Classical classification algorithms were used to classify the direction of the force. As the differences in human users would affect the classification accuracy of this system, there were two human users involved in this research to verify the robustness of system. Experimental results shows that the SVM algorithm can achieve 97.2% and 97.6% recognition accuracy if the experimental data is from the two users respectively. The method could still achieve 96.8% recognition accuracy when the experimental data from the two users together.