MIT News AI reports FLEX helps robots infer higher-level intent from natural-language commands

MIT News AI reports that MIT and New York University researchers have introduced FLEX, a flexible feature-editing pipeline for robot motion planning. The system lets a robot follow commands literally while also using GPT-4o to infer broader intent, such as treating “Don’t spill on the laptop” as a cue to avoid electronics. In simulation, the team tested coffee-serving, candle-moving and bowling-ball tasks, and the full three-step process takes about five seconds.