Google’s robotics unit Intrinsic open-sources its foundational infrastructure for intelligent robots

Google's robotics unit Intrinsic has announced the open-sourcing of its foundational infrastructure for intelligent robots, called Intrinsic Core. Intrinsic Core is a permissive Apache 2.0 licensed software environment designed to run on local hardware, providing a reusable software layer for physical AI. Physical AI refers to AI systems that use sensors, actuators, and control systems to perceive, reason, act, and learn in real-world environments, bridging the gap between bits and atoms.

This move aims to democratize access to intelligent robotics and accelerate the development of physical AI beyond digital applications. Intrinsic Core includes components such as Intrinsic Control for dynamic trajectory adaptation, Pose Estimation based on Nvidia's FoundationPose for dynamic part detection, motion planning for collision-avoidance path generation, and grasp planning for adaptive robotic grippers.

The software also provides simulation and calibration services, along with Intrinsic-ROS drivers for easier integration with third-party hardware components. To demonstrate the potential of Intrinsic Core, the company has released the Open Machine Tending Solution, an open reference design for AI-powered CNC machine tending tasks, supporting modular customization for various hardware setups.

By providing a customizable starting point for robotics developers, Intrinsic hopes to accelerate automation in the machine building industry and foster the growth of a broader robotics developer ecosystem based on shared, open-source tools.