Google's new AI model can teach robots to throw away trash

economictimes.indiatimes.com:

Google has introduced a new AI model called Robotics Transformer 2 (RT-2) that can teach robots to do real-world tasks, like throwing away trash. This is a big step towards having robots that can help humans in various real-world scenarios. Unlike the AI chatbots we know, robots need to understand the real world and do complex tasks. Teaching them has been time-consuming and expensive because they have to learn from lots of data about different objects and situations, Google revealed in a blogpost. But with RT-2, Google has found a better way. It's like a language model for robots. It can understand text and images from the internet and use that knowledge to tell robots how to do specific actions. What's great about RT-2 is that it makes robots smarter. They can think and make decisions based on what they've learned. For example, RT-2 can see and pick up trash without having to be trained on that exact task. It knows what trash looks like and how to handle it, the company stated,. In the past, robotic systems relied on complex stacks of systems to communicate between high-level reasoning and low-level manipulation, which often complicated the decision-making process.

Navigate large documents with ease: Google Docs introduces line numbers for all users

Google for Startups Accelerator invites applications, focus on AI-first startups

Is Bard chatbot as smart as ChatGPT? A look at Google's AI competitor

However, RT-2 streamlines this process by consolidating tasks into a single model, thereby allowing intricate reasoning and direct output of robot actions. Google says that this innovation has yielded impressive results, with