Glossary

Core terms are listed alphabetically. `(Ch N)` marks the chapter where a term is introduced or discussed in depth. Definitions and chapter mappings will be expanded and verified during writing and QA.

Action chunking: A sequence-policy method that predicts several future actions together as an execution chunk. (Ch 5)

Admittance control: A contact-control method that maps measured interaction force into a target motion. (Ch 3)

Behavior cloning (BC): An imitation-learning baseline that learns observation-to-action mappings from demonstrations with supervised learning. (Ch 5)

Constraint projection: The stage that maps a policy or operator action into feasible joint, workspace, collision, and force constraints. (Ch 8)

Diffusion policy: A policy family that generates action trajectories through conditional denoising. (Ch 6)

Flow matching: A method that learns a velocity field between distributions to generate action samples. (Ch 6)

Impedance control: A control method that specifies a target dynamic relationship between motion error and interaction force. (Ch 3)

Receding horizon: Predict several actions, execute only the leading portion, and replan from a new observation. (Ch 6)

Residual policy: A learned bounded correction added to a validated nominal controller or trajectory. (Ch 7)

Safety supervisor: An independent layer that monitors execution layers and manages motion authority, limits, watchdogs, and fallback. (Ch 1)

Teleoperation: A closed loop that transforms human input into robot commands through explicit frame, time, limit, and feedback contracts. (Ch 4)

Trajectory generation: The stage that turns a geometric path into an executable trajectory with velocity, acceleration, timing, and controller references. (Ch 2)

Vision-language-action model (VLA): A model that connects visual and language inputs to robot actions or action representations; detailed treatment belongs to #S13. (Ch 1)

World model: A learned environment transition or latent dynamics model used for prediction and planning; detailed treatment belongs to #S13. (Ch 1)