Paxini Gen3 Tutorial Index
Step-by-step guides from first contact data to tactile-aware robot learning policies. Start with Tutorial 1 if you have just completed the software setup. Jump to Tutorial 3 or 4 if you already have a working sensor stream.
Basic Sensing: First Contact Reading
Install the Paxini SDK, connect the sensor over USB-C, and write a 5-line Python script that prints live pressure data. Run the built-in heatmap visualizer to confirm every taxel is working. Complete this before any other tutorial.
Understanding Tactile Data
Learn the spatial structure of the pressure array, calibrate contact detection thresholds for your gripper, implement a grasp detection function, and visualize contact events over time. Covers the difference between pressure, force, and contact area.
Syncing Tactile with a Robot Arm
Mount the Gen3 on your gripper, route USB cables along the arm, and use MultiSourceSync to record synchronized (arm state, tactile, camera) frames. Verify timestamp alignment and record a 10-second test episode.
Recording a LeRobot Tactile Dataset
Use the extended LeRobot dataset schema to record 50 grasp-and-place demonstrations with tactile channels. Apply the quality checklist, flag slip events automatically, and verify your dataset is ready for policy training.
Training a Tactile-Aware Policy
Add observation.tactile as an input modality to ACT or Diffusion Policy. Compute normalization statistics from your dataset, run training, and evaluate against a vision-only baseline on deformable and transparent objects.
Real-Time Grasp Quality Detection
Implement a lightweight online classifier that runs at sensor rate (500 Hz) and flags unstable grasps in real time. Use the output as a reward signal for online policy adaptation.
Prefer a structured learning path?
Tutorials 1–5 above are the same units in the official Paxini Gen3 Learning Path — a sequenced 8-hour curriculum with completion checks, time estimates, and progress tracking. Start there if you are new to tactile sensing.