Dynamic modelling of multi-physical domain system by bond graph approach and its control

using flatness based controller with MATLAB Simulink

Dr. Shital S. Chiddarwar

Supervisor

Robotics Lab

Saumya Ranjan Sahoo

Research Scholar

Robotics Lab

Department of Mechanical Engineering

Visvesvaraya National Institute of Technology, Nagpur

Need of reliable modelling and simulation

tool for multi domain physical system

Development of Simulink environment

for multi domain system

Controller design to achieve desired

task for multi domain system

Real Time and hardware implementation

Development of a mathematical model for

Mobile robot using Bond graph technique

Design of the bond graph model of the

Mobile robot using MATLAB Simulink

Design of a Flatness based controller to

achieve desired trajectory

Real time implementation of the

proposed

Methodology using MATLAB hardware

support package

Visvesvaraya National Institute of Technology, Nagpur

Approach used to solve the problem

Visvesvaraya National Institute of Technology, Nagpur

Modeling

Controller designOptimization

Real Time Implementation

Modelling Challenges & Solutions With MATLAB/SIMULINK

Visvesvaraya National Institute of Technology, Nagpur

BG V2.0 Tool Box

Functionblocks

SIMULINK

SolverIN

SIMULINK

Kinect supportpackage

Arduinosupport

package

M-file

Dynamic Modelling using Bond Graph

Non-Linear controller model

Solving Second order non linear Equations

Feedback data collection

Real Time Hardware Implementation

Modelling Tool Used

Visvesvaraya National Institute of Technology, Nagpur

SIMULINK

BG V2.0 Tool box

Parameter estimation

Tool box

Kinect Support

Package

Arduino

Support package

Bond Graph

Bond Graph

For

Multidomain

system

Energy

flow(Effort/

Flow)

Bond Graph

Elements

Bond Graph

Junction

Global

Structure

Visvesvaraya National Institute of Technology, Nagpur

BG_V20 Library in SIMULINK

Four wheel omnidirectional mobile robot (FWOMR)

Motor

Direction of

motion

(FWOMR)

Goal

Desired path

Path 1

Path 3

Application

Warehouse Wheel chair Omni move lifter Home Purpose Robots

Visvesvaraya National Institute of Technology, Nagpur

Kinematic modelling of the robot

Visvesvaraya National Institute of Technology, Nagpur

Dynamic modelling using Bond graph of the robot

Bond Graph Of

the Robot

DC Motor Gear Box

Mecanum

Wheel

Translational

MotionYaw Motion

Visvesvaraya National Institute of Technology, Nagpur

Kinematic and Dynamic Model Using Simulink

Visvesvaraya National Institute of Technology, Nagpur

Kinematics Dynamics

Flatness based Differential Control

3

2

1

3333

3333

3

2

1

3333

3333

3

2

1

00

02

)sincos(2

2

)sin(cos2

02

)sin(cos2

2

)sin(cos2

100

02/)sincos(2/)sin(cos

02/)sin(cos2/)sin(cos

y

y

y

X

I

m

yy

m

yym

yy

m

yy

y

y

y

Xyyyy

yyyy

u

u

u

3938373

3625242

1312111

3

2

1

.

.

.

yyy

ref

yyy

ref

yyy

ref

ekdtekeky

ekdtekeky

ekdtekeky

y

y

y

Where

333

222

111

yye

yye

yye

refy

refy

refy

The Controller gain used for the simulations is

calculated by MATLAB Parameter estimation

tool box

Visvesvaraya National Institute of Technology, Nagpur

Flatness Based Controller in

Simulink

Parameter Estimation toolbox For optimized controller parameter

Visvesvaraya National Institute of Technology, Nagpur

Real Time Implementation of the proposed Scheme

Implementation of control algorithm

Simulink model for arduino microcontroller

Robot with arduino microcontroller

Visvesvaraya National Institute of Technology, Nagpur

Real Time Implementation of the proposed Scheme Feedback data collection using Kinect sensor

Visvesvaraya National Institute of Technology, Nagpur

Dynamic model and control

BG+FBC

Kinect based tracking

Reference Trajectory

(In form of flat output)

Computed control input

is send from

SIMULINK to

Arduino board

Robot performance in the

Experimental Environment

+

-

Experimental Environment

Visvesvaraya National Institute of Technology, Nagpur

Computed control input

To robot

Result Achieved

Video

Visvesvaraya National Institute of Technology, Nagpur

Real Time Tracking of the Robot Using MATLAB

Result Achieved (More complicated trajectory)

Video

Visvesvaraya National Institute of Technology, Nagpur

Conclusion

❖Bond graph technique proved to efficient to model complex

Multi-physical system. BGV_20 is found superior to other

software like 20sim, CAMP-G etc.

❖Flatness based controller essentially helps to linearize the system

to get control law.

❖Estimation of controller parameters using optimization tool box

was efficiently done when used with simulation.

❖Real time implementation of designed approach was successful

due to hardware support packages.

Visvesvaraya National Institute of Technology, Nagpur

Why MATLAB ?

Provides all basic and

advance tool for

Scientific computation

and research

Effective Tools for

model based design

of multi domain

system

Implementation of

control strategy

using different

blocks in

SIMULINK

Estimation of

System

ParametersAbility to

conduct number

of simulations

and experiments

Wide range of

Hardware support

(like arduino,

Kinect and others)

Real Time

Implementation