Research Analyzer
← Back ICRA 2026

A Data-Driven Approach for Control and Stabilization of a Single Actuator Monocopter

Danial Sufiyan, Luke Soe Thura Win, Shane Kyi Hla Win, Tee Meng Tan, Shaohui Foong

PDF

AI summary

Key figure (auto-extracted from paper)
A data-driven neural stabilizer enables reliable autonomous outdoor flight for a mechanically simple single-actuator monocopter without external motion capture.
Single-actuator monocopter Neural network stabilization IMU state estimation Autonomous UAV control Data-driven flight control Outdoor MAV deployment

Problem

Monocopters offer mechanical simplicity but are notoriously difficult to stabilize outdoors because their constant rotation makes direct orientation estimation from onboard IMUs unreliable, forcing prior work to rely on controlled indoor motion capture.

Approach

The authors train a lightweight 1D-convolutional neural network to predict world-frame angular velocity directly from raw IMU and velocity data, feeding this estimate into a proportional stabilizer that runs alongside standard P/PID controllers.

Key results

  • 9% prediction accuracy improvement via data augmentation
  • Real-time deployment on an 85g microcontroller with 17 ms inference at 25 Hz
  • Median position errors of 0.5 m, 1.05 m, and 2.22 m for position hold, waypoint, and continuous tracking
  • Successful autonomous semi-outdoor flight in 1.5 m/s winds without external orientation references

Why it matters

Demonstrates a practical pathway for deploying simple, low-cost, and mechanically robust UAVs in real-world outdoor environments where GPS or motion capture is unavailable.

Abstract

In this paper, we use a machine learning approach to stabilize a Single Actuator Monocopter (SAM), showing its ability to operate autonomously outdoors utilizing an onboard Inertial Measurement Unit (IMU). We introduce a neural network-based proportional stabilizer that works in parallel to cascaded P/PID controllers. This network uses the IMU’s data to predict the world frame angular velocity, which is then used to stabilize the SAM. Training data was collected to establish correspondences between the IMU readings and the world frame angular velocity from flights conducted within an indoor motion capture environment. We used data augmen- tation to improve the network’s generalization and prediction performance by 9%. Once trained, the neural network was deployed on the SAM to estimate its angular velocity in real time. We then tested the SAM’s autonomous capabilities in a large semi-outdoor space of approximately 16,000 m3 with wind disturbances of up to 1.5 m/s. We demonstrate position hold, waypoint, and continuous tracking tests, achieving median position errors of 0.5 m, 1.05 m, and 2.22 m, respectively, where no stabilization would result in failure of the defined tests.

Index terms

Aerial Systems: Mechanics and Control Aerial Systems: Applications

Related papers