VIMPPI: Enhancing Model Predictive Path Integral Control with Variational Integration for Underactuated Systems

2025-05-07

AI Summary

This work addresses the challenge of simultaneously achieving limited planning horizons and real-time performance for underactuated double-pendulum systems in the AI Olympics competition. We propose a Variational Integration-Enhanced Model Predictive Path Integral (VIMPPI) control method, which is the first to embed variational integration into the MPPI framework—extending the effective prediction horizon significantly without increasing computational complexity. By integrating control interpolation with online disturbance detection, VIMPPI achieves long-horizon stable prediction and strong robustness at high control frequencies (500–700 Hz). Experimental results demonstrate that VIMPPI substantially outperforms baseline controllers and existing MPPI variants on competition tasks, while achieving both millisecond-level real-time responsiveness and high-precision trajectory tracking—a dual breakthrough in performance and efficiency.

Technology Category

Humans and AI: Human-Aware Planning and Behavior PredictionIntelligent Robots: Motion and Path PlanningMultiagent Systems: Multiagent Planning

Application Category

Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applicationsResponsible Web: Machine-in-the-loop, human agency and autonomyEconomics, Online Markets and Human Computation: Research challenges in human and human-AI computation

Abstract

This paper presents VIMPPI, a novel control approach for underactuated double pendulum systems developed for the AI Olympics competition. We enhance the Model Predictive Path Integral framework by incorporating variational integration techniques, enabling longer planning horizons without additional computational cost. Operating at 500-700 Hz with control interpolation and disturbance detection mechanisms, VIMPPI substantially outperforms both baseline methods and alternative MPPI implementations

Problem

• Enhancing control for underactuated double pendulum systems
• Extending planning horizons without computational overhead
• Improving performance with high-frequency disturbance detection

Innovation

• Enhances MPPI with variational integration
• Operates at 500-700 Hz efficiently
• Includes control interpolation and disturbance detection