It looks like Google will be using Gemini 1.5 Pro’s ‘long context window.’
After introducing Gemini AI, Google seems to be training its robots with Gemini AI 1.5 Pro. Reportedly, Google plans to make robots get better at navigation and completing tasks with Gemini AI.
According to a new research paper, Google’s DeepMind robotics team explained how they are training robots to navigate and complete tasks using Gemini 1.5 Pro. It looks like Google will be using Gemini 1.5 Pro’s ‘long context window.’ Google seems to be planning to mark a significant step forward for AI-assisted robots.
Decoding ‘DeepMind’
Google explained that the robot can then undertake commands based on what it has observed using verbal and / or image outputs. For example, guiding users to a power outlet after being shown a phone and asked “where can I charge this?”
DeepMind also claimed that its Gemini-powered robot had a 90 percent success rate across over 50 user instructions that were given in a 9,000-plus-square-foot operating area.
But what does ‘long context window’ mean and how can it make robots better. So, this “long context window” lets the AI process and understand vast amounts of information in one go. It has the potential to be a game-changer for how robots can learn about and interact with their environment. Given below is the key highlights of how the process works :
- Researchers film a tour of a place, like an office or home.
- The robot, powered by Gemini 1.5 Pro, watches this video.
- The robot learns the layout, where things are, and key features of the space.
- When given a command later, the robot uses its “memory” of the video to navigate.
Can Robots be as smart as humans?
Researchers also found ‘preliminary evidence’ where Gemini 1.5 Pro enabled its droids to plan how to fulfill instructions beyond just navigation. For example, “ a user with lots of Coke cans on their desk asked the droid if their favorite drink is available,”. In response to the instruction, Gemini should be able to help the robot navigate to the fridge, inspect if there are Cokes, and then return to the user to report the result,the team explained.
The robot took around 10-30 seconds to identify the object. The research team thinks that it can be too slow for the real world. In addition to this DeepMind says it plans to investigate these results further.
Furthermore, they are also working on making the system faster and able to handle more complex tasks. Moreover, as the AI models improve we might eventually have robots that understand and move almost like humans do.
Source:financialexpress.com