Back to Projects
ICRA Workshop Paper Xi'an, China

Real-Time MPC with Mobility-Enhanced Feature

Real-Time Control Mobility Maximization Legged Robots NMPC

A real-time NMPC scheme incorporating a mobility-based cost function to significantly improve leg workspace margin and terrain adaptation.

Project Overview

This work introduces a Real-Time Nonlinear Model Predictive Control (NMPC) specifically designed for legged robots to achieve dynamic locomotion across challenging and unstructured environments.

Key Technical Highlights:

  • Mobility-Based Criterion: Maximizes leg joint mobility margins to prevent singularity and enhance step clearance over obstacles.
  • Real-Time Execution: Enables online re-planning at 25 Hz over a 2-second prediction horizon.
  • Efficient Formulation: Employs a single rigid body dynamic model to guarantee computational performance.
Read Workshop PDF