What it is
Sim-to-real is not one trick. It includes domain randomization, calibration, residual learning, hardware tests, and failure logging.
Why it matters
It is the bridge between cheap iteration and physical execution.
How not to overread it
Transfer remains context-specific and needs measured evidence.
Related edges
Simulation
Training to hardware
Transfer requires real-world measurement.
Safety evaluation
Real-world validation
Simulation success is not safety clearance.