Fuzzy neural network based learning control of unmanned aerial vehicle


To develop a controller with in-flight learning capability for a Y6 coaxial tricopter UAV.


The performance of a model-based controller is satisfactory of having a precise mathematical model of the system as well as negligible uncertainties and parameter changes.  However, for the case of small size UAVs, it is a time consuming and expensive task to obtain a precise model of the system. Moreover, such small UAVs are more vulnerable to transient environment conditions, which could lead to control instability.

In such cases, a model-based controller with learning capability or an adaptive model-free controller is preferable. The most significant advantage of a controller with learning capability is that it is able to expand the operating envelope that can not be modelled experimentally. A Type-2 fuzzy neural network controller has been implemented for position control in ROS environment as well as in actual flight tests. Preliminary tests have been run in Gazebo simulator to ensure the functionality of the proposed controllers under different simulated wind conditions and noise level. With better understanding of the control dynamic gained from the simulation tests, the implementation has been extended to actual flight tests using an in-house developed micro Y6 tricopter UAV. In the presence of wind disturbance of 5.5 m/s, the overall result indicated that the tracking accuracy delivered by fuzzy neural network controllers is nearly 35% higher than the conventional PD controller.

                            Schematic of the fuzzy logic control system for a quadcopter UAV

                                                         Reference tracking performance of PD controller and the hybridization of PD                                                                     and Type-2-Fuzzy-Neural-Network (T2-FNN) in the presence of wind.


  1. Brune Cowan, Nursultan Imanberdiyev, Changhong Fu, Yiqun Ding, Erdal Kayacan, "A Performance Evaluation of Detectors and Descriptors for UAV Visual Tracking". 2016 International Conference on Control, Automation, Robotics and Vision (ICARCV2016).
  2. Nursultan Imanberdiyev, Changhong Fu, Erdal Kayacan and I-Ming Chen, "Autonomous Navigation of UAV by Using Real-Time Model-Based Reinforcement Learning". 2016 International Conference on Control, Automation, Robotics and Vision (ICARCV2016).
  3. Erdal Kayacan, Reinaldo Maslim, "Type-2 Fuzzy Logic Trajectory Tracking Control of Quadrotor VTOL Aircraft with Elliptic Membership Functions". Mechatronics, IEEE/ASME Transaction on, Sep., 2016.
  4. Changhong Fu, Andriy Sarabakha, Erdal Kayacan, Christian Wagner, Robert John and Jonathan, "A Comparative Study on the Control of Quadcopter UAVs by using Singleton and Non-Singleton Fuzzy Logic Controllers". 2016 IEEE International Conference on Fuzzy System (FUZZ-IEEE2016).

Principal Investigators

Assistant Prof. Erdal Kayacan (NTU)

Telephone: 6790 5585
Office: N3.2-02-28
Mr. Paul Tan (STE)

Telephone: 6660 1052
Office: S1-B4a-03