Tag: sim-to-real transfer

A technical survey finds that reliable physical AI depends on co-designing learned world models, uncertainty representation, prediction horizons and closed-loop ...

A reinforcement-learning method combining contraction theory was tested in simulation and on three robot platforms, providing stability guarantees and improving ...

A new gradient-decoupled privileged baseline method lifts simulated drone navigation success to 96.47 percent under simultaneous wind, noise, delay and ...

A systematic study shows that carefully engineered synthetic orchard imagery, refined to match real annotation geometry and mixed with a ...

A new reinforcement learning framework called Skill-Nav uses waypoints as a simple interface to combine agile quadrupedal locomotion with general ...

A new systematic review maps how embodied intelligence, from vision-language-action models to world models, is transforming robotic manipulation and what ...

Researchers have built a vectorized simulation and reward framework that trains autonomous microrobot navigation policies in under ten minutes, enabling ...

A systematic review of 191 studies finds that AI methods are transforming robotic peg-in-hole assembly, but direct experimental evidence for ...

A comprehensive new review maps how reinforcement learning, from Q-learning to deep multimodal frameworks, is transforming robotic peg-in-hole insertion for ...