Google’s DeepMind team is making strides in robotics with the help of Gemini AI, specifically the Gemini 1.5 Pro model.
By leveraging this advanced AI, Google aims to enhance its robots' ability to navigate and complete tasks more efficiently.
The magic happens through video tours of designated areas, like homes or office spaces. Researchers use Gemini 1.5 Pro to make the robots “watch” these videos, enabling them to learn about their environment. This allows the robots to follow natural language instructions, such as guiding users to a power outlet when shown a phone and asked, “where can I charge this?”
DeepMind’s research paper highlights that their Gemini-powered robots achieved a 90 percent success rate across over 50 user instructions in a 9,000-plus-square-foot area. This impressive feat showcases the potential of AI in practical applications.
Want to learn more about AI's impact on the world in general and property in particular? Join us on our next Webinar! Click here to register
But it doesn’t stop at navigation. Preliminary evidence suggests that Gemini 1.5 Pro can help robots plan tasks beyond just moving around. For example, if a user asks if their favourite drink is available, the robot can navigate to the fridge, check for the drink, and report back. This level of interaction could revolutionise how we use robots in daily life.
While the video demonstrations are impressive, it’s worth noting that the robots take between 10–30 seconds to process each instruction. So, while we’re not quite at the stage of having ultra-quick, environment-mapping robots in our homes, the progress is promising.
Imagine a future where your robot can find your missing keys or wallets with ease! For now, we’ll keep an eye on these developments and look forward to a smarter, more efficient robotic helper.
Want to learn more about AI's impact on the world in general and property in particular? Join us on our next Webinar! Click here to register
Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai
コメント