[Paper Review] A Real-time Control Approach for Unmanned Aerial Vehicles using Brain-computer Interface

R. Vishwanath, Saumya Kumaar Saksena+1 more|arXiv (Cornell University)|Sep 2, 2018

EEG and Brain-Computer Interfaces12 references4 citations

TL;DR

This paper proposes a real-time brain-computer interface (BCI) control system for fixed-wing and VTOL unmanned aerial vehicles (UAVs) using EEG signals, combining Common Spatial Pattern (CSP) and Linear Discriminant Analysis (LDA) for high-speed, accurate classification at 97 Hz. The method achieves 85% classification accuracy for 2-class control and up to 92% for 4-class control on a multi-rotor UAV, demonstrating stable, mind-controlled flight with low latency and high reliability.

ABSTRACT

Brain-computer interfacing (BCI) is a technology that is almost four decades old and it was developed solely for the purpose of developing and enhancing the impact of neuroprosthetics. However, in the recent times, with the commercialization of non-invasive electroencephalogram (EEG) headsets, the technology has seen a wide variety of applications like home automation, wheelchair control, vehicle steering etc. One of the latest developed applications is the mind-controlled quadrotor unmanned aerial vehicle. These applications, how- ever, do not require a very high-speed response and give satisfactory results when standard classification methods like Support Vector Machine (SVM) and Multi-Layer Perceptron (MLPC). Issues are faced when there is a requirement for high-speed control in the case of fixed-wing unmanned aerial vehicles where such methods are rendered unreliable due to the low speed of classification. Such an application requires the system to classify data at high speeds in order to retain the con- trollability of the vehicle. This paper proposes a novel method of classification which uses a combination of Common Spatial Paradigm and Linear Discriminant Analysis that provides an improved classification accuracy in real time. A non-linear SVM based classification technique has also been discussed. Further, this paper discusses the implementation of the proposed method on a fixed-wing and VTOL unmanned aerial vehicles.

Motivation & Objective

• To address the challenge of real-time, high-speed classification of EEG signals for reliable control of fixed-wing and VTOL UAVs.
• To overcome the limitations of standard classifiers like SVM and MLP in high-speed control scenarios due to slow processing.
• To develop a robust, low-latency BCI system capable of sustaining stable, continuous control of UAVs using only brain signals.
• To validate the system on both a prototype elevon-controlled delta-wing UAV and an off-the-shelf quadrotor UAV.
• To demonstrate the feasibility of non-invasive BCI for complex aerial vehicles requiring rapid response and high accuracy.

Proposed method

• EEG signals were acquired in real time using the EMOTIV EPOC+ headset at a sampling rate of 250 Hz.
• Signal preprocessing included artifact removal, band-pass filtering (8–30 Hz), and epoching based on event-related potentials (ERPs).
• Feature extraction used the Common Spatial Pattern (CSP) method to enhance signal-to-noise ratio for motor imagery tasks.
• Classification was performed using Linear Discriminant Analysis (LDA) for 2-class control and Non-Linear SVM (NLSVM) for 4-class control.
• The system operated at a real-time processing rate of 90–97 Hz, enabling low-latency command transmission to the UAV.
• Control signals were sent via WiFi to a ground station, which relayed high-level commands to the UAV’s flight controller using sensor fusion via Extended Kalman Filtering (EKF).

Experimental results

Research questions

• RQ1 Can a BCI system achieve real-time, high-speed classification of EEG signals for stable UAV control, overcoming the latency limitations of traditional classifiers like SVM and MLP?
• RQ2 How effective is the combination of CSP and LDA in improving classification accuracy and speed for motor imagery-based BCI in fixed-wing UAVs?
• RQ3 What level of control accuracy and response time can be achieved using non-invasive EEG for 2-class and 4-class UAV control tasks?
• RQ4 How does individual cognitive performance, such as focus duration and mental training (e.g., yoga), affect BCI classification accuracy in UAV control?
• RQ5 Can the proposed BCI framework be successfully extended and validated on both prototype fixed-wing and commercial multi-rotor UAV platforms?

Key findings

• The CSP-LDA classifier achieved a real-time processing rate of 97 Hz, enabling low-latency control suitable for dynamic UAV flight.
• The 2-class LDA-based system achieved an average classification accuracy of 85% across all subjects, with subject 10 reaching 98% due to prior yoga practice.
• The 4-class NLSVM-based system achieved up to 92% accuracy on the quadrotor UAV, demonstrating feasibility for complex control tasks.
• Subject 10’s performance (98% accuracy) was significantly higher, suggesting that mental training like yoga enhances BCI control capability.
• The system demonstrated stable, continuous control of both a prototype elevon-equipped delta-wing UAV and an AR Parrot 2.0 quadrotor, with real-time response and minimal latency.
• Offline analysis showed a 11% error rate in the twin dataset evaluation, confirming model robustness and generalization potential.