[ad_1]
Genesis AI, a startup that aims to build a foundational model for powering all kinds of robots, has emerged from stealth with a giant $105 million seed round co-led by Eclipse Ventures and Khosla Ventures.
Founded last December by Zhou Xian (pictured above, left), who holds a PhD in robotics from Carnegie Mellon University, and Théophile Gervet (pictured above, right), a former research scientist with the French AI lab Mistral, the startup wants to build a general-purpose model that will enable robots to automate a wide range of repetitive tasks, from lab work to housekeeping.
Large language models are trained on vast datasets of text, but AI models for robotics must be trained on data on the physical world. However, acquiring that real-world data makes for a costly and time-consuming endeavor.
To overcome that, Genesis is turning to synthetic data, which it generates using a proprietary physics engine that, it says, is capable of accurately modeling the physical world.
Genesis’ synthetic data engine originated from an academic project that Xian led in collaboration with researchers from 18 universities. Several participants from that project have since joined Genesis, making up its current staff of over 20 researchers who specialize in robotics, machine learning, and graphics.
Genesis claims its proprietary simulation engine allows it to develop models faster, a distinct advantage over competitors who rely on Nvidia’s software.
Other companies working on developing general-purpose AI models for robots include Physical Intelligence, which raised a $400 million round, and Skild AI, which was valued at $4 billion earlier this year.
“It’s a big unknown: Will anybody have a large robotics foundation model that will generalize across tasks? That’s a bet we want to go after,” Kanu Gulati, a partner at Khosla Ventures, told TechCrunch.
“Of all the teams we have seen, we like [Genesis’] approach for going after robotics foundation models,” she added.
Genesis is developing its synthetic data and building the foundational model across two offices, in Silicon Valley and Paris.
As the next milestone, Genesis plans to release its model to the robotics community by the end of the year.
[ad_2]
Source link
