Yamakawa Laboratory, The University of Tokyo

東京大学　山川研究室

Although developed for years and attracting a lot of attention, vision-based tactile sensors still have common defects -- the application of such devices on inferring the direction of external forces is poorly investigated, the operating frequency of these devices is too low for them to be applied in practical scenarios. Moreover, dedicate discussion concerning the deformation of elastomer used in vision-based tactile sensors remains insufficient. This research focuses on analyzing the deformation of the thin elastic layer on a vision-based tactile sensor by establishing a simplified deformation model, which is cross-validated using finite element method. Further, this model instructs the reduction of marker amount of a vision-based tactile sensor prototype -- HiVTac. In subsequent testing, the proposed system demonstrates superior accuracy on reconstructing the external force. The average errors on inferring the magnitude and spatial direction of the external force are respectively 0.008 N and 0.32°. The prototype demonstrates its ability of working at 100 Hz sampling rate and 1.3 kHz processing frequency even on a plain PC, enabling real-time reconstruction of the proposed system towards not only the direction but also magnitude of external force.