Hyper-Tactile’s tactile AI algorithm model serves as the core component within the intelligent closed loop of "Perception – Understanding – Execution". Centered on our self-developed Tactile AI Engine, the model undergoes continuous training leveraging edge-side models and a data acquisition center. Rigorous testing is carried out across three dimensions: accuracy, response speed and stability. It efficiently converts collected multi-dimensional tactile signals (pressure, temperature, sliding, touch, etc.) into executable, generalizable, end-to-end tactile data.
触觉AI算法模型
It converts tactile signals into machine-interpretable tactile data for robots, supporting pose recognition, hardness recognition, shape recognition, curved surface recognition, texture recognition and temperature recognition.
The edge-side model efficiently translates tactile signals into executable motion strategies to enable real-time responses by robots.
Relying on the data acquisition center in our in-house laboratory, full data backflow and adversarial algorithm training are implemented to continuously iterate the model and enhance its generalization capability.
