SA-MPPI: Sensitivity-Aware Model Predictive Path Integral Control for Robust and Agile Quadrotor Flight
Fuqiang Gu, Xu Lu, Huidong Liu, Jiangshan Ai, Xianlei Long, Tao Jiang, Chao Chen, Zhao Huang
AI summary
Problem
Standard sampling-based MPC degrades under unmodeled disturbances and uncertainties, leading to brittle control or excessive computational costs that hinder real-time quadrotor agility.
Approach
The framework decouples high-frequency control from a slower auxiliary controller to inform sampling, while penalizing high-variance trajectories via a novel open-loop sensitivity metric evaluated through nested Monte Carlo rollouts.
Key results
- Reduces tracking errors by up to 47% under significant wind disturbances
- Achieves over 2× higher computational efficiency with fewer trajectory samples
- Successfully tracks aggressive trajectories in real-world experiments with strong fan-induced winds
- Decouples real-time control from slower optimization to preserve responsiveness without sacrificing convergence
Why it matters
Provides a practical, low-latency control solution for autonomous drones operating safely in unpredictable, real-world environments.
Abstract
Reliable quadrotor control in dynamic environ- ments remains challenging due to external disturbances and internal uncertainties. While Model Predictive Path Integral (MPPI) control enables agile maneuvers through sampling- based optimization, its performance often degrades under such unmodeled uncertainties, leading to brittle and unsafe behavior. To address this, we propose SA-MPPI, a novel robust MPC framework that integrates asynchronous guidance with a novel open-loop sensitivity metric. The asynchronous module leverages a slower auxiliary controller to generate an informed sampling distribution, improving convergence without intro- ducing latency. The sensitivity metric penalizes high-variance trajectories under sampled disturbances via nested Monte Carlo rollouts, embedding robustness directly into the optimization. Extensive simulations and real-world quadrotor experiments demonstrate that SA-MPPI outperforms adaptive baselines, reducing tracking errors by up to 47% under significant wind disturbances while achieving over 2× higher computational efficiency. These results highlight SA-MPPI’s ability to deliver low-latency, safe, and predictable control in uncertain, dynamic environments.