Apple’s AMP1 system enables general-purpose AI models to control robot arms by writing and executing Python code, without training specific to the robot or task. In simulation, AMP1 achieved an 85.6% success rate with Gemini 4 Argon and 82.8% with Claude Opus 5.5, compared with 74.8% for NVIDIA’s Astranomy model. Researchers also conducted 108 real-world trials across three Franka robot arm configurations, where AMP1 completed 104 trials, for a 96.3% success rate, versus 49 and 30 trials for the baseline systems. AMP1 reduced operational failures by 15 to 20 times, mainly through improved error recovery. The code and simulation data were made public, allowing other laboratories to test the results.
Source: Read the original article
Disclaimer: this content is for information purposes only and is not financial advice. Cryptocurrencies are highly volatile: you may lose all of your capital. Always do your own research. Legal notice
