Coming out of stealth, Tera AI has announced that it has raised $7.8 million in a seed round to develop software known as zero-shot navigation for autonomous hardware-like robots. The funding, co-led by Felicis and Inovia Capital, includes participation from Caltech + Wilson Hill, and Naval Ravikant.
San Francisco-based Tera AI was founded by Tony Zhang, who previously led machine learning at Google X, and now serves as company CEL. Tera AI has a team of AI and simulation researchers from Google AI, Caltech, MIT, and the European Space Agency.
“Spatial reasoning is one of those innate skills humans have that’s really underappreciated and missing in foundation models today,” Zhang said. “Most robots today lack the ability to move between physical environments, and 99 percent of autonomous systems use little to no AI for navigation. Humans can easily navigate between indoor and outdoor places with just our eyes.”
Zhang said people’s brains passively builds internal models of the outside world, which they can reuse years later to recognize exactly where we are. “We can read a map we’ve never seen and use it to navigate a maze-like indoor space. At Tera AI, we believe zero-shot navigation for robots is a problem that software-alone can solve,” he said.
Tera said it will use the funding to deploy an initial solution into embedded devices later this year and will be growing the technical team across both product and research.
Zhang told TechCrunch that new capabilities for existing robots in autonomy will be more affordable with Tera AI’s solution. Localization sensors, lidar system, a pricey Inertial Measurement Unit or even a high-precision GPS receiver can be cost prohibitive for a lighter robot that costs less than $50,000.
