By Scaled Foundations / September 27, 2023

Hello World: Launch Of Scaled Foundations’ Robotics AI Development Platform

The General Robot Intelligence Development (GRID) platform combines foundation models, simulation, and large language models for rapid prototyping

The General Robot Intelligence Development (GRID) platform combines foundation models, simulation, and large language models for rapid prototyping of AI capabilities in robotics

 

We founded Scaled Foundations to help companies across sectors build AI-powered robotic capabilities and applications for critical tasks with ease.

Societally, the need for robotic automation has never been greater, with deglobalization of manufacturing supply chains, aging infrastructure, climate change, and significant labor shortages across industries.

While the promise of robotics and autonomy has been evident for some time, programming robots remains complex and costly, requiring dedicated teams of robotics engineers and proprietary systems, often resulting in techniques that struggle to generalize. Further, integrating AI with robotics has proved challenging due to limited availability of comprehensive robotic data sets.

Today, we’re excited to launch the Pre-Release version (download and try here) of our General Robotic Intelligence Development (GRID) platform which aims to enable rapid prototyping of AI capabilities for robots. GRID is comprised of a curated set of AI foundation models for robotics, synthetic data generation via its high-fidelity simulation engine, and an LLM-based toolchain for orchestration, training, and fine tuning.

Our integrated approach allows non-technical users to simulate specific operating environments and robot physics quickly and accurately, train robots to perform tasks in various operating conditions, infuse AI capabilities for safe and intelligent operations, and export this code onto robotic fleets for real world missions.

Like the promise of LLM-powered Copilots for application development, our value to customers is to significantly reduce development time, cost and expertise associated with creating and deploying advanced robotic solutions.

Initially, we are focusing on aerial robotics and vehicles, with a cloud-native offering coming later this year. New robotics categories, such as manipulation and wheeled robots, will be enabled in the future.

A whitepaper outlining our vision can be found here.

We’re thankful to have support on this journey from our investment partners, Khosla Ventures, the first venture capital backer of Open AI, and E14, MIT’s venture fund.