$pi$-MPPI: A Projection-based Model Predictive Path Integral Scheme for Smooth Optimal Control of Fixed-Wing Aerial Vehicles

2025-04-15

AI Summary

To address the non-smooth control sequences generated by the Model Predictive Path Integral (MPPI) algorithm in nonlinear Model Predictive Control (MPC) for fixed-wing aircraft—which often induce system oscillations—this paper proposes an enhanced MPPI framework incorporating a differentiable projection filter (π-filter). The π-filter enforces hard constraints on both the control inputs and their arbitrary-order derivatives directly within the sampling loop, replacing conventional post-hoc smoothing. This design ensures strict differentiability and full compatibility with end-to-end neural-accelerated optimization. Experimental validation on a real fixed-wing platform demonstrates significantly smoother control outputs, markedly improved robustness, simplified parameter tuning, and low computational overhead. Moreover, the π-filter is modular and plug-and-play, readily integrable into any existing MPPI-based control system without architectural modification.

Technology Category

• Search and Optimization: Mixed Discrete/Continuous Search
• Planning, Routing, and Scheduling: Mixed Discrete/Continuous Planning
• Intelligent Robots: State Estimation

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Model Predictive Path Integral (MPPI) is a popular sampling-based Model Predictive Control (MPC) algorithm for nonlinear systems. It optimizes trajectories by sampling control sequences and averaging them. However, a key issue with MPPI is the non-smoothness of the optimal control sequence, leading to oscillations in systems like fixed-wing aerial vehicles (FWVs). Existing solutions use post-hoc smoothing, which fails to bound control derivatives. This paper introduces a new approach: we add a projection filter $pi$ to minimally correct control samples, ensuring bounds on control magnitude and higher-order derivatives. The filtered samples are then averaged using MPPI, leading to our $pi$-MPPI approach. We minimize computational overhead by using a neural accelerated custom optimizer for the projection filter. $pi$-MPPI offers a simple way to achieve arbitrary smoothness in control sequences. While we focus on FWVs, this projection filter can be integrated into any MPPI pipeline. Applied to FWVs, $pi$-MPPI is easier to tune than the baseline, resulting in smoother, more robust performance.

Problem

• Ensures smooth control sequences for fixed-wing aerial vehicles
• Bounds control derivatives without post-hoc smoothing
• Minimizes computational overhead with neural-accelerated optimization

Innovation

• Projection filter ensures bounded control derivatives
• Neural accelerated optimizer reduces computational overhead
• Integrates smoothly into any MPPI control pipeline