Google DeepMind has unveiled two advanced artificial intelligence models designed to allow robots to perform human tasks without training.
Gemini Robotics and Gemini Robotics-ER give robots spatial understanding and reasoning abilities in order to carry out physical actions autonomously.
A series of video demonstrations showed robots equipped with the new technology carrying out real-world tasks like preparing a packed lunch, tying shoe laces and tidying up objects.
The AI is also capable of performing skills that have been unique to humans up until now, such as folding origami models and figuring out how to “slam dunk” a basketball, without ever having seen a basketball or net before.
“In order for AI to be useful and helpful to people in the physical realm, they have to demonstrate “embodied” reasoning – the humanlike ability to comprehend and react to the world around us – as well as safely take action to get things done,” said Carolina Parada, who heads Google DeepMind’s robotics division.
“Both of these models enable a variety of robots to perform a wider range of real-world tasks than ever before... We look forward to exploring our models’ capabilities and continuing to develop them on the path to real-world applications.”
As well as physical tasks, the new AI comes equipped with a moral compass in an effort to limit risks with the new technology.
The data-driven framework is designed to “steer a robots behaviour” to act in the best interests of humans.
Inspired by Isaac Asimov’s Three Laws of Robotics, the ASIMOV model is the most complex set of rules for robots that has ever been deployed in working machines, according to Google DeepMind.
“This framework would allow people to create, modify and apply constitutions to develop robots that are safer and more aligned with human values,” said Dr Parada.
“Finally, the new ASIMOV dataset will help researchers to rigorously measure the safety implications of robotic actions in real-world scenarios.”
The new AI models are currently only available to “trusted testers”, which include Boston Dynamics and Agile Robots, with the hope that they will be developed for use in commercial machines.
