TacPulse is apulse-sensing system built on vision- based tactile technology. Pulse information is conventionally acquired using optical or electrical techniques, often relying on electrode patches or ultrasound imaging. In contrast, TacPulse employs a bio-inspired, fingertip-sized soft sensor that concentrates pressure over the radial or ulnar artery, while an embedded camera captures skin deformations in real time. From frame-to-frame structural similarity, we establish a method to capture micrometre-scale deformations; a concise and efficient signal-processing pipeline yields high-SNR pulse waveforms. Across varying contact conditions, we observed reproducible systolic-diastolic dual-peak morphology of peripheral pulses, with amplitude ratios and normalized intervals consistent with qualitative expectations of peripheral pulse morphology. Furthermore, we report substantial signal-to-noise ratio gains and qualitative morphology-level correspondence with canonical cardiac waveform features. To our knowledge, this is the first vision-based tactile system to clearly and reproducibly capture dual-peak pulse morphology, enabling quantitative morphology analysis. These results suggest that vision-based tactile fingertips could provide a promising route for future low-cost physiological sensing and human-centred robotics applications.
