Back to Projects
Robotics & Optimization MDPI Robotics (2023)

Optimization-Based Reference Generator for NMPC

Legged Robots LIP Models NMPC Reference Generation Disturbance Rejection

A novel optimization-based reference generator using a Linear Inverted Pendulum (LIP) model to calculate optimal reference trajectories for NMPC, eliminating heuristic tuning.

Project Overview

Model Predictive Control (MPC) approaches often rely on heuristic references and careful tuning. This project introduces an optimization-based reference generator that leverages a Linear Inverted Pendulum (LIP) model to calculate reference trajectories for the center of mass automatically.

Key Benefits & Findings:

  • Automated Tuning: Eliminates manual heuristic tuning for reference trajectory calculations.
  • Improved Disturbance Rejection: Enhances push recovery performance when encountering significant environmental disturbances.
  • Validated Performance: Tested extensively through simulations and real hardware experiments on quadruped platforms.
Read Paper (MDPI) Watch Demonstration