Robot control
Preview control
Definition
Preview control uses a known or predicted future reference over a finite horizon when calculating the current control action. Humanoid walking controllers have used it to shape center-of-mass motion from a preview of the desired zero-moment-point trajectory.
Also known as: Preview controller
Updated
Act now using a future reference
A feedback controller normally reacts to the current error. A preview controller also receives upcoming reference values or disturbances. It can begin changing the system before a future transition instead of waiting for that transition to create an error.
In the classic Kajita and colleagues walking method, a preview of the desired zero-moment-point sequence is used to generate a biped's center-of-mass trajectory. The formulation is tied to a simplified walking model and a stated preview horizon.
Walking references extend across steps
A planned footstep sequence implies future support regions. Preview control lets the current center-of-mass command account for several of those future references, which can produce a smoother approach to the next support transition than tracking only the present reference.
Preview control and model predictive control both look ahead, but they are not interchangeable names. A linear preview controller can use precomputed gains for a fixed model and horizon. MPC usually solves a finite-horizon optimization again online and can include explicit state, input, or contact constraints.
Prediction quality and horizon matter
The controller cannot anticipate an event that is absent from its preview. If a foot lands away from its planned location, the original future reference may no longer match the physical support. State feedback and replanning are still needed.
A short horizon may react too late to a coming transition, while a longer horizon increases dependence on future references and model assumptions. The common walking formulation inherits limitations of the linear inverted pendulum model, including simplified height and angular-momentum behavior. Preview tracking alone does not enforce reachability, friction, collision, or actuator limits.
Sources
Related terms
Zero-moment point
The zero-moment point is a point on a chosen support plane where the net moment associated with the ground reaction wrench has zero components parallel to that plane. In flat-ground walking with the usual contact assumptions, it coincides with the center of pressure.
Linear inverted pendulum model
The linear inverted pendulum model approximates a walking robot by a mass moving at constant height above its support, with simplified angular-momentum dynamics. These assumptions make horizontal center-of-mass acceleration linear in the displacement from the support point.
Model predictive control
Model predictive control repeatedly optimizes future actions using a system model, applies the next part of the solution, and replans from updated state information. It can account for objectives and constraints over a finite prediction horizon.
Bipedal locomotion
Bipedal locomotion is movement using two legs, with body motion coordinated through changing contacts between the feet and the environment. It includes walking and running.