Google has introduced a new artificial intelligence model named Robotics Transformer 2 (RT-2), aimed at aiding robots in understanding tasks like trash disposal.
As detailed in a blog post by Google, the newly developed AI model, referred to as Robotics Transformer 2 (RT-2), will serve as a vision-language-action (VLA) model.
It will undergo training using data and images collected from the internet, which it will then translate into actionable tasks for robots.
Training Robots with AI
Although tasks such as garbage collection might appear simple to humans, for a robot to learn, it requires understanding a series of steps.
For example, the robot needs to first recognize which items are considered trash, then learn the method of picking them up and disposing of them. Rather than being programmed to perform these specific tasks, the RT-2 provides the robot with the ability to utilize information from the web to grasp how to carry out the task, even without explicit training on the exact procedures.
Google has announced that the new model significantly enhances the performance of robots in unfamiliar situations, nearly doubling their effectiveness compared to the model’s previous version.
Furthermore, Google noted that the latest version possesses the ability to apply basic reasoning in responding to user commands. However, Google currently does not plan to commercially release or sell robots equipped with this technology on a large scale.
Despite this, there is a growing trend towards the use of robots in settings such as warehouses or as home assistants.
You may also like this content
- Zoom’s AI Discussion and Update
- YouTube’s New Policy on AI-Generated Content Removal
- YouTube’s AI Tools: Dream Screen, AI Dubbing
-
-
-
-
-
-
-
-
-
-
