Google DeepMind installed a system called Gemini1.5 Pro on the robot. This is no ordinary upgrade. This thing gives the robot the ability to remember and navigate.
Imagine that this robot can perform 57 different tasks on an area of nearly 9000 square feet with a success rate of 90%. This is not a simple task. For example, finding a place to paint. The robot not only understands it, but also takes you to a big whiteboard. This operation is simply more reliable than a real person.
The power of this system is that it can handle multimodal long context windows, which means that the robot can not only remember key locations, but also understand human instructions, video guides, and even reason with common sense. Like the example of the Google employee, the robot not only understands the “place to draw”, but also knows to find a place with a large whiteboard.
Moreover, these robots have become familiar with the office environment in previous projects, and they have learned about the spatial layout through “multimodal command navigation demonstrations.” DeepMind’s team also used a layered vision-language-motion (VLA) technology to allow the robot to understand written, drawing commands, and gestural instructions.
The core of this system is that it allows robots to move freely in complex spaces without requiring constant guidance from humans. They can remember the environment, understand instructions, and then complete tasks in their own way. This ability makes robots more flexible and useful in practical applications.
In short, Google DeepMind’s technology not only makes robots smarter, but also allows them to better serve humans in the real world. This is like opening a new door for robots to enter our lives and become partners in our work and exploration of the world. Robots in the future may no longer be cold machines, but intelligent partners in our lives.
If you want to learn more, you can click on the link below the video.
Thank you for watching this video. If you like it, please subscribe and like it. thank
Original English:https://www.theverge.com/2024/7/11/24196402/google-deepmind-gemini-1-5-pro-robot-navigation
Oil tubing: