We Need a Unicode for Tactile Data — Here's Why I Built One

Description and use

TLabel unifies tactile sensor data into a single schema, eliminating the need for sensor‑specific adapters and speeding up model development.

Problem

Tactile sensor data is fragmented across dozens of incompatible formats, forcing developers to write separate adapters for each sensor and hindering progress in embodied AI.

Outcome

By adopting TLabel, developers can load data from multiple tactile sensors into a single, consistent structure, eliminating adapter duplication and accelerating model training and data analysis.

Workflow

• Install TLabel via pip
• Use tlabel.load() to read sensor files (e.g., GelSight, BioTac, PaXini)
• Access the unified TLabelData structure for downstream tasks
• Export data to standard formats (Zarr, LeRobot) or use CLI for quality scoring

Evidence

TLabel is at v0.20.1 with 13 sensor adapters already built. It supports export to FTP-1 Zarr (for foundation models) and LeRobot format.

Source

pip install tlabel

Source