Feedback Linearization

Definition

Feedback linearization is a model-based control technique that cancels a system’s known nonlinear dynamics by inverting them in the control law, leaving a linear closed loop on which simpler (e.g. PD) compensators act. In dambrosio2024redundant it is the inner controller for a free-flying (fully-actuated 6-DOF base) chaser with a 7-DOF arm: the full coupled inertia and convective terms are inverted, and two PD loops — one on the base, one on the manipulator — drive the linearized error. (The source writes the inertia matrix ; the wiki’s canonical symbol is .)

Key Equations

Coupled equation of motion (canonical inertia , convective/Coriolis ):

Model-based feedback-linearization control law with base/arm PD loops:

gloss: pre-multiplying the PD command by and adding back cancels the plant, so each subsystem tracks a decoupled linear error (base held at the synchronized state; arm follows the guidance policy).

Source Support

  • dambrosio2024redundant — the explicit free-flying multibody EoM (Eq. 1) and the model-based feedback-linearization control law with separate base and manipulator PD loops (Eq. 8), used as the inner loop beneath a DRL guidance policy.

Open Questions

  • dambrosio2024redundant reports nonzero PD steady-state error as the positioning bottleneck; does feedback linearization degrade near manipulator singularities where is ill-conditioned, and how does that interact with the redundant arm?