Google has launched its AI model Gemini, which uses large language models (LLM) to give robotic arms instructions in the physical world.
MIT Technology Review reports that the model can follow commands such as “put the grapes in the clear glass bowl”, with the LLM filtering intentions and translating these into robotic commands.
The move follows recent AI/robotic developments from the US-based Open AI and the Chinese model, Manus, as well as Amazon’s acquisition of Covariant.
However, home-based robots will be a later development, as unpredictable home environments will require extensive testing before robots are market-ready.
For now, testing is taking place in warehouses and industrial settings under the supervision of human operators.