Skip to content

dronefreak/system-identification-helicopter

BranchesTags

Repository files navigation

System Identification for Unmanned Helicopter

MATLAB License Status

Parameter optimization for a TREX 550 flybarless helicopter using metaheuristic algorithms and state-space modeling

Table of Contents

About

This project implements system identification for an unmanned TREX 550 flybarless helicopter using the state-space structure based on Bernard Mettler's Yamaha RMAX helicopter model. The system optimizes 40 aerodynamic and control parameters to maximize the correlation between actual flight test data and simulated model outputs.

Objective

Maximize the Pearson correlation coefficient between:

  • Actual flight test data from the TREX 550 helicopter
  • Simulated outputs from the state-space model

The optimization problem involves a 40-dimensional parameter space representing the A, B, C, and D matrices of the state-space model with 13 states, 4 inputs, and 10 measurable outputs.

Features

  • Multiple Optimization Algorithms: Six different metaheuristic algorithms implemented
  • Real Flight Data: Validated using actual flight test data from TREX 550 helicopter
  • State-Space Modeling: Based on Bernard Mettler's proven helicopter dynamics model
  • Performance Optimizations: Parallel computing, memory optimization, and progress tracking
  • Profiling Tools: Identify bottlenecks and optimize execution time
  • 3D Visualization: Advanced animation and visualization capabilities
  • Comprehensive Results: Detailed comparison and statistical analysis tools
  • Modular Design: Easy to extend with new algorithms or cost functions
  • Comprehensive Testing: 47+ test cases covering unit, integration, and regression tests

Installation

1. Clone the Repository

git clone https://github.com/dronefreak/system-identification-helicopter.git
cd system-identification-helicopter

2. Verify MATLAB Installation

Ensure you have MATLAB R2014b or later installed with the required toolboxes (see Prerequisites).

3. Add to MATLAB Path

Open MATLAB and add the project directory to your path:

addpath(genpath('/path/to/system-identification-helicopter'));
savepath;

Prerequisites

Software Requirements

  • MATLAB: R2014b or later (tested on R2014b, R2016a, R2018b)
  • Operating System: Windows, macOS, or Linux

Required MATLAB Toolboxes

  • Control System Toolbox - For state-space models (ss(), lsim())
  • Statistics and Machine Learning Toolbox - For correlation functions (recommended)

Optional MATLAB Toolboxes

  • Parallel Computing Toolbox - For parallel cost function evaluation (2-8x speedup)

Verify Toolbox Installation

% Check if toolboxes are installed
ver('control')
ver('stats')
ver('distcomp')  % Parallel Computing Toolbox (optional)

Third-Party Dependencies

The following third-party libraries are included in the repository:

  • Yarpiz YPEA Framework - Optimization algorithms (BSD License)
  • STLRead - 3D model import (File Exchange)
  • MatlabPlaneGraphics - 3D visualization framework

Quick Start

Run IWO Optimization (Recommended)

% 1. Navigate to IWO directory
cd src/algorithms/iwo/IWO

% 2. Load the workspace with flight data
load('../../../../data/experiments/best.mat')

% 3. Run the optimization (5000 iterations)
iwo

% 4. Visualize results
cd ../../../..
addpath('src/utils')
PopulCheck

Expected runtime: 30-60 minutes (depends on your CPU)

Algorithms Implemented

Algorithm Status Directory Iterations Population Size
Invasive Weed Optimization (IWO) ✅ Tested src/algorithms/iwo/ 5000 40
Genetic Algorithm (GA) ✅ Tested src/algorithms/ga/ 1000 Variable
Artificial Bee Colony (ABC) ⚠️ Experimental src/algorithms/abc/ Configurable Variable
Bees Behavior Algorithm (BBA) ⚠️ Experimental src/algorithms/bba/ Configurable Variable
Particle Swarm Optimization (PSO) 📝 Example src/algorithms/pso_example.m - -
Simulated Annealing (SA) 🔨 In Development src/algorithms/sa/ Configurable Variable

Legend: ✅ Fully tested | ⚠️ Experimental | 📝 Example code | 🔨 Under development

Usage Guide

Running IWO

Invasive Weed Optimization is the recommended algorithm for this problem.

% Step 1: Load the workspace
cd src/algorithms/iwo/IWO
load('../../../../data/experiments/best.mat')  % Contains: inr, outr, t, population

% Step 2: Run optimization
iwo  % Runs for 5000 iterations

% Step 3: Extract best parameters
bestParameters = BestSol.Position;

% Step 4: Visualize results
cd ../../../..
addpath('src/utils')
popul = bestParameters';  % Copy to popul variable
PopulCheck  % Generates comparison plots

What happens during optimization:

  • The Sphere.m cost function creates a state-space model using the 40 parameters
  • Simulates helicopter response using lsim() with actual input data
  • Compares simulated output with actual flight test data
  • Returns a cost value (to be minimized) based on correlation and least squares error

Running Genetic Algorithm

cd src/algorithms/ga
MainCode  % Run genetic algorithm optimization

Running Other Algorithms

Artificial Bee Colony:

cd src/algorithms/abc
main  % Run ABC optimization

Bees Behavior Algorithm:

cd src/algorithms/bba
Main  % Run BBA optimization

Simulated Annealing:

cd 'src/algorithms/sa/01 TSP using SA (Standard)'
% Configure and run

Performance Optimization

The IWO algorithm includes several performance optimization features:

Enable Parallel Computing

For 2-8x speedup with Parallel Computing Toolbox:

% Edit src/algorithms/iwo/IWO/config_iwo.m:
config.useParallel = true;  % Enable parallel evaluation

% Then run normally
load('data/experiments/best.mat')
iwo

Profile Performance

Identify bottlenecks and optimize:

% Profile 100 iterations
load('data/experiments/best.mat')
results = profile_iwo('iterations', 100);

% View detailed profiler report
profview(0, results.profileInfo);

Configure Performance Settings

Check system capabilities and configure:

% Interactive performance configuration
config = configure_performance();

% Disable progress bar for batch processing
config = configure_performance('progress', false);

Performance Features:

  • Parallel Computing: Evaluate cost functions simultaneously (2-8x speedup)
  • Memory Optimization: Preallocated arrays reduce overhead (enabled by default)
  • Progress Tracking: Real-time progress bar with ETA
  • Profiling Tools: Identify and optimize bottlenecks

See docs/PERFORMANCE.md for the complete performance guide.

Parameter Description

The optimization identifies 40 parameters in the state-space model:

State-Space Model Structure

ẋ = Ax + Bu
y = Cx + Du

Where:

  • States (13): u, v, w, p, q, r, φ, θ, ψ, a, b, rfb, c

    • u, v, w: Velocity components
    • p, q, r: Angular rates
    • φ, θ, ψ: Euler angles
    • a, b, c: Rotor flapping angles
    • rfb: Rotor feedback

  • Inputs (4): δlat, δlon, δped, δcol

    • Lateral cyclic, Longitudinal cyclic, Pedal, Collective

  • Outputs (10): Measurable states (excluding rfb, c, d)

Parameter Ranges

Parameter Type Min Value Max Value Description
Velocity derivatives -60 +60 Xu, Xa, Yv, Yb, etc.
Rotational dynamics -220 +220 Angular rate coefficients
Time constants 0 220 Tf, Ts
Control gains -100 +100 Control effectiveness

Total: 40 parameters optimized simultaneously

Project Structure

system-identification-helicopter/
├── src/                                  # Source code
│   ├── algorithms/                       # Optimization algorithms
│   │   ├── iwo/                         # Invasive Weed Optimization
│   │   │   └── IWO/
│   │   │       ├── iwo.m                # Main IWO algorithm
│   │   │       ├── Sphere.m             # Cost function
│   │   │       └── config_iwo.m         # Configuration
│   │   ├── ga/                          # Genetic Algorithm
│   │   ├── abc/                         # Artificial Bee Colony
│   │   ├── bba/                         # Bees Behavior Algorithm
│   │   ├── sa/                          # Simulated Annealing
│   │   └── pso_example.m                # PSO example
│   │
│   ├── models/                          # Helicopter dynamics models
│   ├── utils/                           # Utility functions
│   │   ├── PopulCheck.m                 # Results visualization
│   │   ├── profile_iwo.m                # Performance profiling
│   │   └── configure_performance.m      # Performance configuration
│   └── visualization/                   # 3D visualization & animation
│       └── Animations/                  # Animation framework
│
├── data/                                # All data files
│   ├── flight_data/                     # Raw flight test data
│   │   └── raw/                         # .bin and .mat files
│   ├── experiments/                     # Experimental results
│   │   ├── best.mat                     # Primary dataset
│   │   ├── best2.mat                    # Alternative dataset
│   │   └── *.mat                        # Other results
│   └── logs/                            # Flight and simulation logs
│       ├── flight_logs/                 # 17 flight sessions
│       └── reference/                   # Reference logs
│
├── docs/                                # Documentation & papers
│   ├── PERFORMANCE.md                   # Performance optimization guide
│   ├── IFAC_heli_weed.doc              # Academic paper
│   └── figures/                         # Plots and diagrams
│
├── examples/                            # Example scripts
├── tests/                               # Test suite (47+ test cases)
│   ├── test_sphere_cost.m              # Cost function unit tests
│   ├── test_config_iwo.m               # Configuration tests
│   ├── test_popul_check.m              # Visualization tests
│   ├── test_iwo_integration.m          # Integration tests
│   ├── test_regression.m               # Regression tests
│   ├── validate_installation.m         # Installation validator
│   └── run_all_tests.m                 # Test runner
├── results/                             # Output directory for new results
│
├── README.md                            # This file
├── CONTRIBUTING.md                      # Contribution guidelines
├── LICENSE                              # Project license
├── CHANGELOG.md                         # Version history
└── CITATION.cff                         # Citation information

Note: Each directory contains its own README.md with detailed information.

Results Visualization

Using PopulCheck.m

After optimization completes, visualize the identified parameters:

% Copy best parameters to popul variable
popul = BestSol.Position';

% Run visualization
PopulCheck

This generates:

  • Time-domain comparison plots (actual vs. simulated)
  • State trajectories for all 10 outputs
  • Correlation coefficients for each output
  • Model fit quality metrics

Accessing Raw Results

% Best solution structure
BestSol.Position  % 40 optimized parameters
BestSol.Cost      % Final cost value

% Convergence history
figure;
semilogy(BestCosts);
xlabel('Iteration');
ylabel('Best Cost');
title('Optimization Convergence');

Troubleshooting

Common Issues

1. NaN Values During Initialization

Problem: Random initialization sometimes produces NaN cost values.

Solution: The code uses pre-defined initial populations in best.mat to avoid this issue. If you modify initialization:

% Check for NaN before evaluation
if any(isnan(pop(i).Position))
    pop(i).Position = unifrnd(VarMin, VarMax);
end

2. Missing Toolbox Error

Problem: Undefined function 'ss' for input arguments of type 'double'

Solution: Install Control System Toolbox:

% Check installation
ver('control')

% Or use MATLAB Add-On Explorer to install

3. File Not Found Errors

Problem: Unable to read file 'best.mat'

Solution: Ensure you're in the correct directory:

cd 'YPEA119 Invasive Weed Optimization/IWO'
ls  % Verify best.mat exists

4. Out of Memory Errors

Problem: MATLAB runs out of memory during optimization

Solution: Reduce population size or iterations:

% In iwo.m, modify:
MaxIt = 1000;   % Instead of 5000
nPop = 20;      % Instead of 40

5. Slow Performance

Problem: Optimization takes too long

Solution: Use built-in performance optimization features:

% Enable parallel computing (2-8x faster)
% Edit src/algorithms/iwo/IWO/config_iwo.m:
config.useParallel = true;

% Profile to find bottlenecks
load('data/experiments/best.mat')
results = profile_iwo('iterations', 100);

% Check configuration
configure_performance();

See docs/PERFORMANCE.md for complete performance optimization guide.

Quick tips:

  • Close unnecessary MATLAB figures during optimization
  • Reduce iteration count for testing: config.maxIterations = 1000
  • Monitor progress with built-in progress bar

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

How to Contribute

  1. Fork the repository
  2. Create a feature branch (git checkout -b feature/amazing-feature)
  3. Commit your changes (git commit -m 'Add amazing feature')
  4. Push to the branch (git push origin feature/amazing-feature)
  5. Open a Pull Request

Areas for Contribution

  • Additional optimization algorithms
  • Performance improvements
  • Better visualization tools
  • Documentation improvements
  • Test coverage
  • Bug fixes

Citation

If you use this code in your research, please cite:

@software{krishnan2019sysid,
  author = {Krishnan, Navaneeth and Kumaar, Saumya},
  title = {System Identification for Unmanned Helicopter using Metaheuristic Algorithms},
  year = {2019},
  publisher = {GitHub},
  url = {https://github.com/dronefreak/system-identification-helicopter}
}

Related Publications:

  • Bernard Mettler's helicopter dynamics papers
  • IFAC paper (see IFAC_heli_weed.doc)

Also see CITATION.cff for structured citation data.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Third-Party Licenses

  • Yarpiz YPEA Framework: BSD License (see individual algorithm directories)
  • STLRead: BSD License
  • MatlabPlaneGraphics: See respective directories for licenses

Authors

Creator: Navaneeth Krishnan Maintainer: Saumya Kumaar

Acknowledgments

  • Bernard Mettler for the helicopter state-space model framework
  • Yarpiz team for optimization algorithm implementations
  • Contributors to STLRead and MatlabPlaneGraphics

Support

For questions, issues, or feature requests:

  • Issues: GitHub Issues
  • Email: Contact the maintainer via GitHub profile

Last Updated: 2026-01-06 Version: 2.0.0

About

Code repo for the paper https://arxiv.org/abs/1809.05021

Topics

genetic-algorithm system-identification invasive-weed-optimization trex-550 helicopter-dynamics

Resources

Readme

License

View license

Contributing

Contributing
Activity

Stars

11 stars

Watchers

2 watching

Forks

4 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages