A Data-Driven Approach to Vibrotactile Data Compression

The emerging Internet of Skills that exchanges tactile and other sensorial data, significantly augments traditional multimedia. The increase of data scale and modalities demands for codecs dedicated to these sensorial data. In this paper, we propose a codec for compression of vibrotactile data in the spirit of Weber’s law. To be specific, a companding function is applied to the vibrotactile data, so that the quantisation step of high amplitude is larger than that of low amplitude. The curve of the companding function is optimised through a data-driven approach. To evaluate the performance of the vibrotactile codec in terms of human perceived quality, rigorous subjective tests are conducted. The results demonstrate that 75% compression of vibrotactile data is achieved without perceivable degradation. More importantly, the computational complexity is much lower and the latency performance is superior, compared with other vibrotactile codecs. The computational complexity of the proposed codec is about 1/20 of that of previous codecs, while the time delay is approximately 1/30 of that of previous codec.