Real-time System Identification of Unmanned Aerial Vehicles: A Multi-Network Approach

Puttige, Vishwas; Anavatti, Sreenatha G.

doi:10.4304/jcp.3.7.31-38

Cited by 21 publications

(9 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…3. Therefore, the quadcopter parameters such as the moment of inertia, force, mass, and engine constant are not required, and the complexity of Equations 1 to 6 can be avoided [31], [32].…”

Section: Time and Frequency Domain System Identificationmentioning

confidence: 99%

Neural Network-Based System Identification for Quadcopter Dynamic Modeling: A Review

Pairan¹,

Shamsudin²,

Zulkafli³

2020

JAMEA

View full text Add to dashboard Cite

A quadcopter is a rotorcraft with a simple mechanical construction. It has the same hovering capability similar to the traditional helicopter, but it is easier to maintain. The quadcopter is hard to control due to its unstable system with highly coupled and non-linear dynamics. In order to design a robust control algorithm, it is crucial to obtain a precise quadrotor flight dynamics through system identification approach. System identification is a method of finding the mathematical model of the dynamics system using the input-output data measurement. Neural network (NN) based system identification is excellent alternative modeling because it reduces development costs and time by avoiding governing equations and large aerodynamic database. NN based system identification has successfully identified the quadcopter dynamics with good accuracy. This paper gives an overview of the characteristic of the quadcopter and presents a comprehensive survey of the modeling techniques used to determine the flight dynamics of a quadrotor with a particular focus on NN based system identification method. The presented research works have been classified into different categories such as the first principle modeling, system identification and implementation of NN based system identification in quadcopter platform. Finally, the paper highlights challenges that need to be addressed in developing efficient NN based system identification model for unmanned quadcopter system.

show abstract

“…3. Therefore, the quadcopter parameters such as the moment of inertia, force, mass, and engine constant are not required, and the complexity of Equations 1 to 6 can be avoided [31], [32].…”

Section: Time and Frequency Domain System Identificationmentioning

confidence: 99%

Neural Network-Based System Identification for Quadcopter Dynamic Modeling: A Review

Pairan¹,

Shamsudin²,

Zulkafli³

2020

JAMEA

View full text Add to dashboard Cite

show abstract

“…Based on different sets of boundary constraints from Eq (25)(26)(27)(28)(29), several trajectories are generated by the optimization algorithm and are illustrated in Fig. 7.…”

Section: B Trajectory Generationmentioning

confidence: 99%

Controlling a Quadrotor Carrying a Cable-Suspended Load to Pass Through a Window

Guo

Zha

et al. 2019

J Intell Robot Syst

View full text Add to dashboard Cite

In this paper, we design an optimal control system for a quadrotor to carry a cable-suspended load flying through a window. As the window is narrower than the length of the cable, it is very challenging to design a practical control system to pass through it. Our solution includes a system identification component, a trajectory generation component, and a trajectory tracking control component. The exact dynamic model that usually derived from the first principles is assumed to be unavailable. Instead, a model identification approach is adopted, which relies on a simple but effective low order equivalent system (LOES) to describe the core dynamical characteristics of the system. After being excited by some specifically designed manoeuvres, the unknown parameters in the LOES are obtained by using a frequency based least square estimation algorithm. Based on the estimated LOES, a numerical optimization algorithm is then utilized for aggressive trajectory generation when relevant constraints are given. The generated trajectory can lead to the quadrotor and load system passing through a narrow window with a cascade PD trajectory tracking controller. Finally, a practical flight test based on an Astec Hummingbird quadrotor is demonstrated and the result validates the proposed approach.

show abstract

“…In this effort the ANN topology of the plant model is optimized using the genetic algorithm (GA) to select the most appropriate values of the time window size of the input and output delays in the NARMAX formulation and the size of the hidden layer that minimizes the training and validation error. The concept of the dualnet model has been originally introduced by Puttige and Anavatti [23]. Specifically, both offline and online ANN models in the form of multilayer perceptrons (MLPs) for the desired system are trained.…”

Section: Introductionmentioning

confidence: 99%

Fault compensation by online updating of genetic algorithm-selected neural network model for model predictive control

et al. 2019

View full text Add to dashboard Cite

This paper presents a methodology that combines a dual-net model and the model predictive control (MPC) to compensate degraded system performance caused by slow-paced faults/anomalies. The dual-net model is comprised of an offline and an online artificial neural networks (ANNs) along with a switch that selects one of them for MPC. Through selective online updating of weight parameters, the online ANN is able to accurately capture the fault-induced variations in system dynamics, and can be used for MPC reconfiguration and fault compensation. Specifically, the system dynamics is identified by training a multilayer perceptron (MLP). To improve the model accuracy, a meta-optimization approach based on the genetic algorithm is applied to optimize the MLP hyperparameters and the training algorithm. A dual-thread decision maker is proposed to manage the robust model updating scheme and the dual-net model switch. A case study of numerical simulation using an unmanned quadrotor is undertaken to verify the feasibility of the proposed method to mitigate performance degradation. Salient performance in the response prediction and control, subject to gradually growing anomaly is successfully demonstrated. Quantitatively, the proposed updating model outperforms the offline ANN model and yields 2× and 4× lower errors, respectively, for prediction and control of the system response.

show abstract

Real-time System Identification of Unmanned Aerial Vehicles: A Multi-Network Approach

Cited by 21 publications

References 22 publications

Neural Network-Based System Identification for Quadcopter Dynamic Modeling: A Review

Neural Network-Based System Identification for Quadcopter Dynamic Modeling: A Review

Controlling a Quadrotor Carrying a Cable-Suspended Load to Pass Through a Window

Fault compensation by online updating of genetic algorithm-selected neural network model for model predictive control

Contact Info

Product

Resources

About