Date: March 29, 2023
Covariant, a Bay Area firm that specializes in building AI that helps robots learn how to pick up different objects, has developed an AI-based system that trains network robots to improve their picking capabilities as they learn.
The system has proved invaluable in automating warehouses, which have to deal with an impossibly broad range of sizes, shapes, textures, and colors. Covariant's system has gone deeper into AI than any other robotic company by starting with pure software and pure AI, with the first hires being AI researchers.
The company takes a foundation model approach to learning, allowing it to train a large foundation-generalized model on more data to make the AI more generalized, unlike other AI models that are specific to a single use case.
Covariant is also working on building a foundation for future applications while focusing on picking and placing. The grasping capability or pick and place capability is definitely the first general capability that Covariant is giving robots.
LinkedIn May Soon Launch Its Own Reels Format
LinkedIn, the world's biggest professional social networking platform, tests short-form video content to improve user engagement.
These 3 Google Maps Updates Will Make Summer Travel Breezy
Google Maps released a new set of updates yesterday that aims to make travel more convenient, comfortable, and advanced for its users.
Facebook Has Been Spying On Snapchat & Amazon For Years
Facebook, the social media platform owned by Meta, is facing new charges for breaching encrypted data and spying on apps like Snapchat and Amazon
Pavan Davuluri Becomes The New Head Of Microsoft Windows
In a major step to boost AI development, Microsoft has appointed Pavan Davuluri, an IIT Madras alumni, as Chief of Windows.