planning
Rollback improves three robot tasks by comparing branches from remembered states News
Rollback reports an average 11.21-percentage-point success improvement on three real-robot tasks by branching imagined actions from previously observed states.
Model predictive control: planning ahead, taking one step, then planning again Lesson
Model predictive control repeatedly forecasts a limited future, executes the first action of a chosen plan and replans from fresh observations under explicit constraints.
Monte Carlo tree search: deciding which possibility to explore next Lesson
Monte Carlo tree search is a planning method that builds a tree of possible actions while using sampled outcomes to focus future computation on branches that look both promising and underexplored.
What are world models? Lesson
A world model is an AI system's internal understanding of how an environment works — not just what it sees right now, but what will happen after an action, and what would have happened differently. Central to planning, robotics, and the next generation of physical AI.