isaac protocol

The problem isaac protocol solves Problem

Humanoid robots like Tesla Optimus and Figure AI are coming fast. We want using robots should be as easy as using ChatGPT(mass adoption of robotics). But their biggest bottleneck isn’t hardware, it’s data.

Robots need millions of demonstrations of how humans move, grasp, and interact with the world. Right now, only a handful of labs can collect that data, and it’s slow, expensive, limited and raises privacy concerns.

Solution

We capture motion data using simple webcams without storing any video. Our system verifies(zk proofs) and quantifies data quality through AI agents, enabling transparent, trustworthy datasets for robotics training. We reward the users for their data based on the quality(score, accuracy) of data, so the companies can be sure( or verify) that they are getting quality data. Vision

A data DAO that simplifies how teams collect, distribute, and train robots with spatial data. A plug n play platform, where anyone with an idea can create a personal robot for them. Robots-to-robots zk tunneling to update robots with latest knowledge, without relying on companies to wait for updades(in a safe manner).

We capture user pose using pose detection models from tensorflow and mediapipe. We then convert the 2d data points into 3d using various methods like calculating hand movement and direction. We calculate the score based various factor and reward the users. Score is calculated based on accuracy, timing and obstruction avoidance.