The challenge of solving robotic touch

One post says Astra, Fable and Muse are tackling robotics and world-model benchmarks “zero-shot”—without task-specific examples—and characterizes this as a leap in progress. It also describes sensing despair among academics. A post quoting that assessment emphasizes unfinished work: robotic touch remains unsolved, it argues, and there is little training data, leaving interesting research to do.