← Projects

Soft Robotic Arm Using Isaac Sim

Learning-based control couples a Gaussian-process dynamics model with model-predictive control to track the motions of an Isaac Sim FEM soft arm in real time.

Isaac Sim · FEM Gaussian Process Model Predictive Control Soft robotics

1Demo

GP-MPC tracking on the Isaac Sim soft-arm simulation.

2Overview

The soft robotic arm is controlled with a Gaussian Process-based Model Predictive Control method. The GP model is used to capture the nonlinear soft-arm dynamics, while MPC optimizes the control inputs to track the desired motion in real time.

This simulation shows how Isaac Sim FEM can be combined with learning-based predictive control for soft robotic arm modeling, simulation, and control.

GP-MPC control architecture for the Isaac Sim soft arm

GP-MPC architecture: a Gaussian-Process dynamics model inside the real-time MPC loop.

TECHNICAL SKILLS: Isaac Sim