Dynamics-Invariant Quadrotor Control using Scale-Aware Deep Reinforcement Learning

2025-03-09

AI Summary

To address the robustness challenges in quadrotor trajectory tracking under varying payloads, airflow disturbances, and platform heterogeneity, this paper proposes an end-to-end, force-and-torque–direct deep reinforcement learning (DRL) control framework that enforces physical dynamics invariance. We introduce a novel scale-aware dynamics randomization mechanism parameterized by arm length, enabling zero-shot generalization across a 30 g–2.1 kg mass range. Our architecture integrates a temporal trajectory encoder with a history state-action–driven implicit dynamics encoder, eliminating conventional hierarchical control structures. Extensive real-world closed-loop flight experiments (>200 trials) on the Crazyflie 2.1 platform demonstrate RMSE < 0.05 m at peak speeds of 2.0 m/s, with significantly enhanced resilience to wind gusts, ground effect, and swinging payloads. Tracking accuracy improves by 85% over state-of-the-art DRL baselines.

Technology Category

• Intelligent Robots: Motion and Path Planning
• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Machine Learning: Adversarial Learning & Robustness

Application Category

• 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
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Due to dynamic variations such as changing payload, aerodynamic disturbances, and varying platforms, a robust solution for quadrotor trajectory tracking remains challenging. To address these challenges, we present a deep reinforcement learning (DRL) framework that achieves physical dynamics invariance by directly optimizing force/torque inputs, eliminating the need for traditional intermediate control layers. Our architecture integrates a temporal trajectory encoder, which processes finite-horizon reference positions/velocities, with a latent dynamics encoder trained on historical state-action pairs to model platform-specific characteristics. Additionally, we introduce scale-aware dynamics randomization parameterized by the quadrotor's arm length, enabling our approach to maintain stability across drones spanning from 30g to 2.1kg and outperform other DRL baselines by 85% in tracking accuracy. Extensive real-world validation of our approach on the Crazyflie 2.1 quadrotor, encompassing over 200 flights, demonstrates robust adaptation to wind, ground effects, and swinging payloads while achieving less than 0.05m RMSE at speeds up to 2.0 m/s. This work introduces a universal quadrotor control paradigm that compensates for dynamic discrepancies across varied conditions and scales, paving the way for more resilient aerial systems.

Problem

• Robust quadrotor trajectory tracking under dynamic variations.
• Eliminating traditional control layers using deep reinforcement learning.
• Maintaining stability across drones of varying sizes and conditions.

Innovation

• Deep reinforcement learning optimizes force/torque inputs directly.
• Temporal and latent encoders model platform-specific dynamics.
• Scale-aware dynamics randomization enhances stability across drones.