Applied control theory (nonlinear, adaptive, and robust control systems) Design and automation Dynamic systems

(1)

Activities at the Robotics,

Automation, and

Automation, and Mechatronics

Mechatronics

Automation, and

Automation, and Mechatronics

Mechatronics

(RAM) Laboratory

Dr. Jingang

Dr. Jingang Yi, J g gJ g gYi, ,,Assistant ProfessorAssistant Professorff

Department

Department of Mechanical & Aerospace Engineeringof Mechanical & Aerospace Engineering Rutgers, The State University of New Jersey Rutgers, The State University of New Jersey

Piscataway, NJ 08854 Piscataway, NJ 08854

About Myself

yy



 Education backgroundEducation background

– Ph.D. (ME), UC Berkeley, 2002 – M.A. (Math), UC Berkeley, 2001 – M.Eng., Tsinghua Univ. (China), 1996 – B.S. (EE) Zhejiang Univ. (China), 1993



 RR hh ii 

 Research experienceResearch experience

– Rutgers Univ., 08/2008 -- Now

San Diego State Univ 01/2007 08/2008

– San Diego State Univ., 01/2007-08/2008 – Texas A&M Univ., 01/2005-01/2007 – Lam Research Corp., 05/2002/-01/2005

2

p , / / /

Research

Research Interests

Interests



Research areas

– Applied control theory (nonlinear, adaptive, and

robust control systems)

D i d i

– Design and automation – Dynamic systems



A li

i



Application areas

– Robotics and vehicle systems

Bi di l d bi l i l

– Biomedical and biological systems

– Civil infrastructural and transportation systems

S i d t f t i t

– Semiconductor manufacturing systems

Recent

Recent Research Areas

Research Areas



Autonomous robotic systemsAutonomous robotic systems11::

–– Autonomous robot Autonomous robot and and control control –– Human/robot interaction Human/robot interaction –– Mobile manipulatorsMobile manipulators

Vi i

Vi i bb dd ii ii

–– VisionVision--based navigationbased navigation



Sensing/actuation Sensing/actuation systemssystems22: :

Contact

Contact sensing systemssensing systems

–– Contact Contact sensing systems sensing systems –– NanoNano--/bio/bio--manipulationmanipulation –– Energy harvestersEnergy harvesters



_{Automation science and engineering}_{Automation science and engineering}33

1.

1. ICRA’06ICRA’06--11, 11, IROS’06,07, IROS’06,07, AR’07AR’07, ASME , ASME DSCC’

DSCC’ IEEE TRO’ IEEE TRO’ IEEE TCST’ IEEE TCST’ DSCC’08

DSCC’08--11, IEEE TRO’09, 11, IEEE TCST’11 11, IEEE TRO’09, 11, IEEE TCST’11 2.

2. IEEE IEEE Sensors J’08Sensors J’08--09; SPIE’08, DSCC’08, 09; SPIE’08, DSCC’08, IEEE/ASME TMech.

IEEE/ASME TMech.’08,’08, 3.

(2)

Research Projects

GOALI: Safety-preserved estimation and control of

tire/road interaction (PI NSF) tire/road interaction (PI, NSF)

CAREER: Human-inspired safety-preserved vehicle agile

maneuvers (PI, NSF) maneuvers (PI, NSF)

Automated nondestructive evaluation and rehabilitation

system (ANDERS) for bridge decks (Co-PI, NIST)y ( ) g ( , )

Automated condition assessment of concrete bridge decks

using multiple NDE technologies (Task leader, FHWA)g p g ( , )

BIOME - A bio-robotic infrastructure for oceanic

microbial ecology (Co-PI, NSF)

Some other projects

5

Estimation, Sensing, and Control of

/

d

/

d

Tire/Road Interactions



Tire/road friction estimation and control



In situ

tire/road friction sensing



Stability and agility of aggressive maneuvers

6

Hybrid Tire/Road Friction Models

A hybrid physical/dynamic friction

model

– Partitioned adhesion/sliding regions

(physical model)

Th dh i d f i d f i

– The adhesive deformation and force is

calculated by a LuGre dynamic model

Resolve the limitations of the existing results in literature Bridge the model parameter relationshipsg p p

Enable the use of “smart tire” for parameter estimation

Why Need

Why Need Tire Sensing

yy

Tire Sensing Systems?

gg yy

Systems?



Tire/road friction modeling and estimation

–– Lack of Lack of in situin situsensing informationsensing information –– Tire tread dynamics neglectedTire tread dynamics neglected



Tire pressure monitoring system (TPMS)

–– Only monitors inlet pressure and temperatureOnly monitors inlet pressure and temperature –– TPMS cannot be used in realTPMS cannot be used in real--time applicationstime applications –– Battery lifeBattery life--time limitationtime limitation



Deformation information is critical for

–– Understanding of tire/road interactionsUnderstanding of tire/road interactions

8 –– RealReal--time monitoring and controltime monitoring and control

(3)

Deformation Sensing Systems

g y



Polyvinylidene flouoride (PVDF) sensors



Custom

Custom--built circuits

built circuits

9

Sensor Modeling: Bending



Generated charge due to bending

10



Maximum charge when

Sensing Modeling: Stretch/Compress

g

p



Physical model: a partition

of tire/road contact patch



Strain/stress distribution

on sensors



Maximum charge

Sensing Modeling: Summary

g

y

  Bending+stretch in Bending+stretch in sensor output sensor output sensor output sensor output 

 Adhesion region sizeAdhesion region size



 How to utilize sensorHow to utilize sensor 

 How to utilize sensor How to utilize sensor

reading? reading?

–– Estimate byEstimate by

Estimate sliding friction Estimate sliding friction

–– Estimate sliding friction Estimate sliding friction

coeff. coeff.

(4)

Experimental Setup



On

On--board control systems and a sensor suite

board control systems and a sensor suite

Vi i

b

d

l

li

i

h l b l



Vision

Vision--based computer localization as the global

based computer localization as the global

location reference

Wheel encoders Wheel encoders IMU IMU Control board Control board 13 (a) The experimental robot (b) Various ground conditions (c) Camera positioning systems

C C

Sensor Measurements

Left/right wheels: 60/80 rpms: Concrete surface. Sensors embedded in the right front wheel.

14

Sensor Data Analysis

yy

E i f lidi f i i ffi i

(a) Charge by stretch/compress (b) Estimate of adhesion region size

15

Estimate of sliding friction coefficient on the concrete surface:μ_x=0.47

Embedded Rubber Force Sensors



We design and

fabricate a well

fabricate a

well--structured

structured single

single--tire test bed

tire test bed

t e test bed



Pressure

Pressure--sensitive,

sensitive,

electric conductive

rubber sensor is

embedded inside

the tire



Four sensing cells

16

on one sensor

(5)

Stick

Stick--Slip Tire/Road Interaction Sensing

Slip Tire/Road Interaction Sensing

(a) Vision-based contact patch estimation

 Simultaneously measure both the normal and frictional forces

(b) Contact patch normal load distribution

normal and frictional forces

 A multi-cantilever model for stick-slip

interaction 17

Sensor Calibration and Modeling

(c) Calibration

(a) Sensing schematic and models (a) Sensing schematic and models

(d) Validation

18 (b) Sensor calibration

Frictional Force Estimation

(a) Deformation under varying pulling forces (b) Estimated vs measured friction force distribut

 The multi-cantilever beam model predicts frictional force distribution

on the contact patch during the stick-slip process

(a) Deformation under varying pulling forces (b) Estimated vs. measured friction force distribut.

p g p p

 The model captures the evolution of the stick-slip development over

the contact patch

Estimation, Sensing, and Control of

/

d

/

d

Tire/Road Interactions



In situ

tire/road friction sensing



(6)

Human

Human--Inspired

Inspired Aggressive Maneuvers

Aggressive Maneuvers



 Transfer Transfer humanhuman--inspired driving knowledge to the nextinspired driving knowledge to the

next--generation

generation “accident“accident--free” free” vehicle systemsvehicle systems dd dd dd dd



 Advance Advance understanding and control of humanunderstanding and control of human--machine

machine--environment (HME) interactions environment (HME) interactions



 St bilitSt bilit ddA ilitA ilit ii t t ht t h t i tit i ti ff 

 StabilityStabilityand and AgilityAgilityare important characteristics of are important characteristics of

aggressive maneuvers aggressive maneuvers

21

(a) J

(a) J--turn turn (b) Handbrake turn (b) Handbrake turn

Vehicle Maneuver Agility Metrics



Agility metric 1

: lateral jerk (i.e., acceleration

derivative)

Aggregated over distance



Agility metric 2

: relative lateral acceleration

Aggregated over distance gg g

22

Pendulum

Pendulum--Turn Aggressive Maneuver

Turn Aggressive Maneuver



A high-speed sharp turning strategy used in rally

race driving



Experiments conducted at Ford facilities by

f

i

l

i

d i

F d SUV

professional racing drivers on a Ford SUV

23

(a) Driver inputs

(a) Driver inputs (b) Vehicle position trajectory(b) Vehicle position trajectory

Typical Human Driver vs. Professional Driver

(b) Trajectory under a human driver model (b) Trajectory under a human driver model (a) Trajectory under professional driver

(a) Trajectory under professional driver

Travel time and agility metrics comparison: professional driver vs. typical human driver Travel time and agility metrics comparison: professional driver vs. typical human driver

(7)

Comparison:

pp

Vehicle Maneuver

Stability

(a) Trajectory stability: professional driver (a) Trajectory stability: professional driver

25

(b) Human driver with

(b) Human driver with aa_xmax_xmax=0.5 m/s=0.5 m/s22 _{(c) Human driver with}_{(c) Human driver with aa} xmax xmax=3 m/s=3 m/s22

Summary



Hybrid physical/dynamic friction model captures

the complex tire/road forces in a compact form

p



A pendulum

A pendulum--turn aggressive maneuver example

turn aggressive maneuver example

has been analyzed for stability and agility

y

g y

comparison



Vehicle can be operated in “unstable regions” to

p

g

achieve high agility aggressive maneuvers



A safety boundary concept is proposed for

y

p

p p

designing new autonomous aggressive maneuvers

26

Rider/Bicycle Interactions and “Smart

Bicycle”

Bicycle” Based Rehabilitation

Based Rehabilitation

Bicycle

Bicycle --Based Rehabilitation

Based Rehabilitation

Video source: SnijdersSnijders and and BloemBloem, “Cycling for freezing of gait,” , “Cycling for freezing of gait,” New Engl. J. Med., vol. 362, 2010.

New Engl. J. Med., vol. 362, 2010.

Rutgers “Smart bicycle” systems New Engl. J. Med., vol. 362, 2010.

New Engl. J. Med., vol. 362, 2010.

IMU/Seat Force Sensor

IMU/Seat Force Sensor--based Rider/Bicycle

based Rider/Bicycle

Pose Estimation

Pose Estimation –– Preliminary Results

Preliminary Results

yy

(c) Rider pose estimation

(a) Rider/bicycle modeling schematic

results

(b) Seat force sensor

(8)

Automated Nondestructive Evaluation and Automated Nondestructive Evaluation and Rehabilitation Systems (ANDERS) for Bridge Decks* Rehabilitation Systems (ANDERS) for Bridge Decks*

* Supported by the NIST TIP Award 70NANB10H014 (2010-2015) and FHWA grant (2011-2013).

Bio

Bio--robotic Sampler for Rutgers

robotic Sampler for Rutgers

Underwater Gliders

 _{Development of in-situ onboard}

biological sampler systems for Slocum biological sampler systems for Slocum autonomous underwater gliders

 _{Extended work for modeling,}

localization, planning and control of single- and multi-glider systems

30 * Supported by the NSF Award OCE-1131022 (2011-2014)

Semiconductor Manufacturing

Automation



Micro

Micro--scale

scale

11

: :

–– Mechatronic systems for Mechatronic systems for

semiconductor manufacturing semiconductor manufacturing

22



Meso

Meso--scale

scale

22

::

–– InIn--situ monitoring, diagnosis and situ monitoring, diagnosis and

t l t l control control



Macro

Macro--scale

scale

33

::

Cl l l i d

–– Cluster tool analysis and Cluster tool analysis and

scheduling scheduling

1.

1. Mech.’05, ASME IMECE’04Mech.’05, ASME IMECE’04 ACC’ I TSM’ I ACC’ I TSM’ I

31

2.

2. ACC’04, IEEE TSM’03,05, IEEE ACC’04, IEEE TSM’03,05, IEEE CSM’08

CSM’08 3.

3. WSC’04, ICRA’05,07, CASE’06WSC’04, ICRA’05,07, CASE’06--08, 08, IEEE TSM’06, IEEE

IEEE TSM’06, IEEE TASE’07,08,11 TASE’07,08,11

Acknowledgments



IndividualsIndividuals

–– UC Berkeley:UC Berkeley:Profs. R. Horowitz, M. Profs. R. Horowitz, M. TomizukaTomizuka, K. Hedrick, and , K. Hedrick, and

SS SastrySastry S. S. SastrySastry

–– TAMU:TAMU:Prof. D. Song, Prof. H. Liang, Prof. D. Song, Prof. H. Liang, ProfProf. R. . R. LangariLangari, J. Zhang, , J. Zhang,

C.

C.--Y. Kim, and B. MikaY. Kim, and B. Mika

R tgers

R tgers::Pr f GPr f G ki Pr f K rkh f Pr f Shki Pr f K rkh f Pr f Sh Pr f LiPr f Li

–– RutgersRutgers::Prof. Gucunski, Prof. Kerkhof, Prof. Shan, Prof. Lin, Prof. Gucunski, Prof. Kerkhof, Prof. Shan, Prof. Lin,

Prof. Basily, Prof. Li, and the RAM Lab members Prof. Basily, Prof. Li, and the RAM Lab members

–– Ford R&D:Ford R&D:Drs. E. H. Tseng, J. Lu, W. Liang and D. Drs. E. H. Tseng, J. Lu, W. Liang and D. HrovatHrovat



_Sponsor_{Sponsor agencies}_agencies

–– National Science Foundation (NSF)National Science Foundation (NSF)

R C f Ad d I f d T i

–– Rutgers Center of Advanced Infrastructure and Transportation Rutgers Center of Advanced Infrastructure and Transportation

(CAIT) (CAIT)

–– Ford Research LabsFord Research Labs

T T

32

–– NISTNIST –– FHWAFHWA

(9)

Robotics, Automation, and

Mechatronics

Mechatronics (RAM) Lab

(RAM) Lab

Email:

[email protected]

jgy

g

http://coewww.rutgers.edu/~jgyi

Thank You!