Hybrid Force/Position Control

In contact tasks, splitting directions apart — controlling force where the robot is constrained, position where it is free.

Proposed by Raibert and Craig in 1981. When a robot touches an object, the environment constrains motion along some directions — for example, an eraser wiping a whiteboard cannot push through the board's surface, but can slide freely within it. The method uses a selection matrix S, with diagonal entries of 1 or 0, to split the six spatial directions (defined in a task-specific coordinate frame) into two groups: the directions S selects track a desired position, and the rest (I − S) track a desired contact force; the two outputs are combined and converted into joint torques. It requires knowing the constraint directions in advance, and getting the surface normal wrong can generate very large forces. Impedance control, by contrast, doesn't split directions at all — it specifies a relationship between force and position error along every direction at once. Note that motor manufacturers' ‘hybrid force/position mode’ usually refers to a single joint's MIT-style command format, an unrelated concept.

ExampleErasing a whiteboard: the x and y directions within the board's plane are position-controlled, moving the eraser along a path; the z direction perpendicular to the board is force-controlled, holding a constant pressing force.

Also called

Hybrid Position/Force Control, Hybrid Motion-Force Control

Related

Impedance Control · Force Control · Admittance Control · Task-Space Control · Constant Force Control (Force Tracking) · Contact-rich Manipulation

Sources

Raibert & Craig, Hybrid Position/Force Control of Manipulators, J. Dyn. Sys., Meas., Control 103 (1981)
Modern Robotics 11.6: Hybrid Motion-Force Control