The authors introduce OpenZilo, an open smart‑ring platform that integrates a microphone, an inertial measurement unit (IMU), photoplethysmography, and skin‑temperature sensors. The ring’s software architecture is split into four layers: sensor acquisition, event interpretation, interaction policy, and robot execution—allowing researchers to plug in or swap components.

In a user study, ten participants performed five different commands, 250 trials each, using either speech or gesture. Speech achieved a 90.4 % task‑success rate, while gesture reached 88.8 %. Recognizer accuracy was even higher than the success rate for both modalities. Gaussian hidden‑markov models (HMMs) outperformed support‑vector machines (SVMs) across three vocabulary sizes, though HMMs required more time to score.

When the robot was instructed with a spoken start command, participants who then handed control to gesture showed a shorter median response time than those who continued speaking. Overall, gesture‑based tasks succeeded in 70 % of trials, compared with 64 % for speech.

The study involved a small cohort and repeated, constrained tasks, so the results may not extend to everyday or continuous use. The paper notes that concurrent multi‑sensor acquisition and long‑term battery life were not fully tested.

The OpenZilo platform is released under an open license, enabling the research community to extend and adapt it for human‑centric robot interaction studies.