MPC

Table of contents

1. Typical pipeline
2. Why it matters in manipulation

---

Model Predictive Control (MPC) optimizes control actions over a finite prediction horizon using a dynamics model, then applies the first action and re-plans at the next step.

Typical pipeline

1. Predict future states with a system model.
2. Optimize a cost over the horizon (tracking, constraints, effort).
3. Apply the first control input.
4. Receed and repeat with updated observations.

Why it matters in manipulation

• Handles constraints on joint limits, contact, and workspace
• Useful for contact-rich and tracking tasks
• Bridges trajectory planning and closed-loop control