Rethinking Reference Trajectories in Agile Drone Racing: A Unified Reference-Free Model-Based Controller via MPPI

2025-09-18

AI Summary

In drone racing, conventional model-based controllers rely on precomputed reference trajectories—such as full-state trajectories or geometric paths—limiting time-optimality and online adaptability. This paper proposes a reference-free unified Model Predictive Control (MPC) framework that directly encodes the racing objective as maximization of gate-passing progress, enabling end-to-end time-optimal flight. Inspired by reward shaping in reinforcement learning, we integrate a differentiable, optimizable gate-progress reward function into the Model Predictive Path Integral (MPPI) algorithm. Furthermore, we design a unified manifold to enable fair performance comparison across diverse objective formulations. Experimental results on real-world racing platforms demonstrate that our approach matches or surpasses classical trajectory-tracking and contouring controllers in competitive scenarios, while significantly improving agility and robustness under dynamic constraints and environmental uncertainties.

Technology Category

• Planning, Routing, and Scheduling: Model-Based Reasoning
• Intelligent Robots: Motion and Path Planning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

While model-based controllers have demonstrated remarkable performance in autonomous drone racing, their performance is often constrained by the reliance on pre-computed reference trajectories. Conventional approaches, such as trajectory tracking, demand a dynamically feasible, full-state reference, whereas contouring control relaxes this requirement to a geometric path but still necessitates a reference. Recent advancements in reinforcement learning (RL) have revealed that many model-based controllers optimize surrogate objectives, such as trajectory tracking, rather than the primary racing goal of directly maximizing progress through gates. Inspired by these findings, this work introduces a reference-free method for time-optimal racing by incorporating this gate progress objective, derived from RL reward shaping, directly into the Model Predictive Path Integral (MPPI) formulation. The sampling-based nature of MPPI makes it uniquely capable of optimizing the discontinuous and non-differentiable objective in real-time. We also establish a unified framework that leverages MPPI to systematically and fairly compare three distinct objective functions with a consistent dynamics model and parameter set: classical trajectory tracking, contouring control, and the proposed gate progress objective. We compare the performance of these three objectives when solved via both MPPI and a traditional gradient-based solver. Our results demonstrate that the proposed reference-free approach achieves competitive racing performance, rivaling or exceeding reference-based methods. Videos are available at https://zhaofangguo.github.io/racing_mppi/

Problem

• Eliminates reliance on pre-computed reference trajectories for drone racing
• Optimizes direct gate progress instead of surrogate tracking objectives
• Handles discontinuous objectives via sampling-based MPPI control framework

Innovation

• Uses MPPI for real-time optimization
• Incorporates gate progress objective directly
• Eliminates need for pre-computed references