Google is using its advanced AI model, Gemini 1.5 Pro, to train robots to navigate the real world.
DeepMind, Alphabet’s AI research lab, recently released footage of a robot trained on Gemini 1.5 Pro traversing through its busy office.
According to multiple sources, including The Verge, DeepMind uses the Gemini AI model to train robots using video tours. This helps them learn to move and complete tasks through visual inputs.
Gemini 1.5 Pro can understand massive amounts of data from long videos and documents.
By processing up to 10 million tokens, it handles much more data than previous models.
The AI splits tasks into smaller chunks and assigns them to specialized groups, known as expert networks, that handle specific types of tasks. This method, called Mixture-of-Experts, makes the model very efficient.
#ai #ainews #deepmind #googlegemini #google #artificialintelligence #machinelearning #technews #robotics #robots #humanoidrobot #advancedai #machinelearning #llm
