What is a generalist robot policy?
A generalist policy aims to perform multiple tasks or operate across multiple conditions rather than optimizing one fixed behavior.
Why generalization is hard
Tasks vary in objects, goals, environments and physical interactions. A policy must generalize without losing the low-level reliability required by hardware.
Embodiment as an interface
Generalist behavior becomes more reusable when the policy can separate task-level intent from embodiment-specific execution. This connects generalist policies to cross-embodiment learning and robot foundation models.
Explore the field
Continue through the research map, model directory, robot directory and company directory.