26th International Symposium on Automation and Robotics in Construction (ISARC 2009)

317

Development of a Control System for a Multipurpose Road Repairing Machine

Pekka Kilpeläinen1, Mika Jaakkola2 and Pauli Alanaatu3

1VTT Technical Research Centre of Finland, P.O. BOX 1100, Oulu, FIN-90571, Finland; email: Pekka.Kilpelainen@vtt.fi 2Destia Ltd, P.O.Box 444,Oulu, FIN-90101,Finland;email: mika.jaakkola@destia.fi 3Destia Ltd, Turuntie 207,Espoo, FIN-02740,Finland;email: pauli.alanaatu@destia.fi

Abstract

In this paper an automatic control system for a multipurpose road pavement repairing machine (ROADMOTO) is introduced. ROADMOTO machine is equipped with asphalt milling drum and two asphalt spreaders. The old wearing course of the road is heated and milled. Asphalt spreader in the middle of the machine is used for spreading the crushed old pavement. Asphalt spreader in the back of the machine is used for spreading the new asphalt mass on top of the old layer. Until now most functions of the machine have been manually controlled. The goal is to achieve cost saving and better work quality by using automation. Before the actual repairing work a GPR (Ground penetrating radar), a profilometer or laser scanning techniques are used for collecting information about the road. Designing of repairing tasks is based on the collected data. During a design phase a repair design file is created. ROADMOTO machine is equipped with a GPS positioning unit and the repair design file can be used for automatic control of road repairing operations. The control system also offers a manual control mode as well as automatic height and slope control modes. This ensures flexibility, because the user can choose control mode that best suits for the situation. The control system uses CAN bus as sensor and valve interface and hydraulic actuators are closed loop controlled to achieve high control accuracy. The concept from the data collection and design to the automatic machine control is presented as well as the developed prototype system and results from the first tests. Keywords: Automation, pavement repairing machines, hydraulic control systems

1. Introduction

It has been estimated that as much as 85% of Europe’s road construction projects today include different repairing and rehabilitation operations. Automation is one of the modern means for improving process efficiency and product quality in road construction as well as in road maintenance. The benefits of automation will be produced through the entire construction process. Automated and model based process means exploitation of developed design-, control- and positioning systems in different phases of road construction process.

The process of data flow in road repairing and rehabilitation construction from the automation point of view consists of 1) initial data collection and problem diagnosis, 2) repairing and rehabilitation design, 3) site operations including machine control operations, and 4) quality control actions.

The initial data collection consists of survey methods providing the basic information for rehabilitation design are as follows: a) GPR (ground penetrating radar) for thickness surveys and detecting reasons for damages, b) FWD (falling weight deflectometer) for stiffness measurements of structural layers and subgrade, and c) profilometer or laser scanning techniques to collect data from the road surface. These techniques need accurate positioning systems in order to produce data for precise 3-D road models. The rehabilitation design and machine control models can be processed with special cad tools made for that purpose.

In this paper a control system for a multipurpose road repairing machine (ROADMOTO) is presented (Figure 1). ROADMOTO machine is equipped with asphalt milling drum and two asphalt spreaders. The old wearing course of the road is heated and milled. Asphalt spreader in the middle of the machine is used for

spreading thenew asphalt laasphalt is used

The idea othe worst defthe road geom100% and ovrepairing mod

ROADMOremixer machThe spreader

2. Surveying

Surveying and applying

2.1 Survey and Surveying

system contai

e crushed old ayer on top ofd on the bottoof the machinfects of the rometry is needeverall cost savidel. OTO machinhine. In this ca

in the middle

Figu

and Modelin

and modellinautomation fo

d 3D model of the road is

ins laser, full i

Aut

pavement. Af the old layerom layer. ne is to use oload and to mied about 20 kging is about 1

ne is also equase old crushee of the machi

re 1. Multipur

ng

ng phases (Figor the repairin

Figure 2.

s done by speinertial and G

tomation and Ro

3

Asphalt spreadr. So two layer

ld heated andinimize the usg/m2. It is est5%. In this ca

uipped with aed pavement ine is not used

rpose road rep

gure 2) are imng of the pave

Process to su

ecial measuremGPS devices. S

obot Application

318

er in the backrs of asphalt a

d crushed pavese of new pavtimated that wase cut and fil

a mixing drums mixed with

d in the remixe

pairing machin

mportant whenment.

urvey and repa

ment system mSystem collect

ns

k of the machiare done at the

ement as mucvement. Typicwith automatioll operations a

m and it can the new aspher work.

ne (ROADMO

n designing re

air the road.

mounted in a ts all the data

ine is used foe same time a

ch as possibleally new mass

on this can be are done acco

be used alsoalt mass in the

OTO).

epairing mode

car (Figure 3)from the sen

r spreading thand old crushe

e for correctins for correctinreduced 50%rding to a roa

o as an asphae mixing drum

el for road sit

). Measuremensors to the raw

he ed

ng ng

% - ad

alt m.

te

nt w

data file dlongitudin

Accura

horizontalapplicationthickness i

2.2 Design fDesign

specified. done and and decidedamages n

The ampossible toused for fi

When ROADMOInput datawhole concoordinate

3. Machin

3.1 Main fuThe de

operation - N

sites. - Su

machiThe m

vary on a possible.

The aumiddle of back of th

Both thautomatio

26th Internatio

during the mnal road profile

acy for the 3Dl direction grns. Ground pis needed.

for repair n for repair isIn the designiwhat the targeed how to fix need special acmount of aspo mill more thilling. GPR dadesigning is

OTO. Repairia file defines tnstruction sitee is given and

ne Control Sy

functions evelopment wof the machin

No surveying a

urveying and ine control. Su

machine controwork site con

utomatic contf the machine.he machine is che cutter and

on system con

onal Symposium

measurement. es are combin

D model in verid resolutionpenetrating ra

s done with ing phase of tet of repair is.them. There

ctions when rephalt used is han needed foata is useful in ready the ping file is crethe milling depe. If spreader it is related to

ystem

work of the cone. Two main and modelling

modelling isurveying and mol system shounsiderably, cho

trol system is . These are recontrolled mathe spreader

ntrols are cutt

m on Automati

Next phase ined in to one 3

Figure 3.

ertical direction is possibleadar can be u

terrain modelthe repair proc Different kincan be e.g. edepaired. optimized in

or repairing ann optimizationplan is insereated and ROpth and filling

is used in tho milled surfac

ontrol system situations are

g is done befo

s done beforemodelling are uld be applicaboosing betwee

used for coneferenced beloanually and is nare moved by

ting depth (or

ion and Roboti

319

is the proces3D surface of

Surveying sys

on is providede to choose used, if infor

lling applicaticess it has to bnds of things odge drops, pot

the designinnd there will b

n. rted in to th

OADMOTO’sg height for eahe repairing wce.

started with e: orehand. This

ehand and a typically doneble for both o

en manual and

ntrolling the mow as the cuttnot connectedy two hydraulr height) h1 an

ics in Construc

sing of the cf the road.

tem.

d by high resoto be high

rmation about

ion. Design isbe known whaof the pavemetholes or deep

ng phase. E.g.be sections wh

he automated control softw

ach grid coordwork, the spre

analysing diff

s is typical situ

repairing moe only in mainof these situatid automatic op

milling drum ater and the spd to the automlic cylinders (Fnd h2 and slop

ction (ISARC 2

collected data

olution laser uenough for t road structu

s done with at kind of repent condition p rutting. Var

. there are sehere the crush

d pavement rware can read

dinate. The inpeader height l

ferent working

uation in sma

odel is availan roads and larions. Because peration shou

and the asphapreader. Asphmation systemFigure 4). Thepe θ of the cu

009)

a. Transversal

units (0.1 mm)terrain mode

ure e.g. pave

the accuracy airing work whas to be survrious kinds of

ections where hed asphalt w

repairing macd it as input put data file colevel for each

g situations du

all roads and w

able for automrge work sitesthe circumsta

uld be as flexib

alt spreader inhalt spreader in. e variables thautter (or sprea

l and

). In elling ment

level will be veyed f road

it is will be

chine data.

overs h grid

uring

work

matic s. ances ble as

n the n the

at the ader).

Automation and Robot Applications

320

Ultrasonic sensors are used for measuring the actual height h1m and h2m relative to the road surface and inclination sensor is used for measuring actual slope αm.

Figure 4. Hydraulic cylinders of the cutter and the spreader

The ROADMOTO machine weights 40 tons, length is 16 m and wheelbase is about 8.5 m. The machine

is equipped with rubber tyres in the front and air tyres on the back. Compression of the tyres and the torsion of the frame also affect the position and orientation of the cutter and the spreader. Operating modes

To allow most flexibility to the user, the control system was designed so that user can set a control mode individually for each of the four cylinders. Different control modes are marked by letters J, U, K and M. Control modes are:

- J (Joystick mode): Full manual control using a joystick. - U (Ultrasonic mode): Cutting depth and height control using ultrasonic sensors (Figure 4 number 1). - K (Slope control): Automatic slope control using an inclination sensor (Figure 4 number 2). - M (Model control): Automatic slope control according to the repairing model. Changing between control modes can be done on-the-fly during operation. Because the user can choose

from four different modes for both cylinders, there are 16 possible combinations of control modes. Although the number of combinations is quite high, the control logic is easy to learn and offers flexibility.

3.2 Control method

Controlling the motions of the cutter and the spreader, a Cartesian control method is used (Figure 5). The benefit of the Cartesian control is that different Cartesian values can be controlled independently (height of left end h1 and height of the right end h2 and absolute slope α) according to the selected control mode (m1 and m2). Joysticks (j1 and j2) can override automatic control.

Figure 5. Cartesian control method.

Outputs from the Cartesian control are Cartesian velocities •

1h , •

2h and •

θ . From these set values for the positions (l1 and l2) and velocities (v1 and v2) of the cylinders are calculated. Positions of the cylinders are then closed-loop-controlled. Status of the actuator control is used as feedback. If delay of cylinders exceeds limits, Cartesian velocities are limited.

In this

kinematicsdemandingcontrol bagrader [Ki

3.3 HardwThe co

system comodules, such as ulprotocol (bus reduce

Controby the usecontrol tainterface performan

User inmodes andinterface.

3.4 Model bIn the

(Real Timrepairing msystem viaare contro

4. Prototy

The symonths. Imodelling

4.1 Test sitTest si

approximatraffic.

Conditedge drop

Repairimounted repair wor

26th Internatio

case the mecs gives more g and high acased on real tilpeläinen 199

ware and softwareontrol system onsists of twowhich are coltrasonic senso(e.g. directionaes the amountol software is er interface msks. The benmodule is ba

nce for contro

nterface modd set values (e

based control momodel based

me Kinematic)model accorda CAN bus. Ifolled automatic

ype Test

ystem was takeIn October 20

to the autom

te ite is typical ately 2000 veh

tion of the tess, but only 20ing work witand raw datark were succes

onal Symposium

chanism that iadvances wh

ccuracy is neetime positioni9], stabilizatio

e architecture i

o user interfaconnected togeors, are conneal valves and it of wiring conimplemented

module. Both nefit of this is ased on a 16

ol tasks.

dule consists oe.g. height can

ode control mode

) GPS receivding to the pof the user hascally accordin

en in use in Ju008 a test wa

matic machine c

road in southicles. There w

st site was goo0 – 30 cm fromth ROADMOa was collectessful. The test

m on Automati

is controlled ien controllingded. Some aping and 3D m

on cutter [Kilp

is a combinatice modules (other using a ected to the Cinclination sennsiderably. in a centralizof the user inthe duplicatio

6 bit digital

Figure 6. Usof joystick fo

n be set in 1 m

e the ROADMver. PC composition of macs selected modng to the value

une 2008. Duris arranged in control.

th Finland (Fwere no proble

od and the mam the edge waOTO was dond during the

t site considere

ion and Roboti

321

is quite simpleg more comppplications aremodels of thepeläinen 2004]

ion of centralone in both sCAN bus (C

CAN bus via nsors) are con

ed way, whichnterface moduon of control signal proces

ser interface mor manual con

mm steps). Cur

MOTO machputer handles chine. These vdel control mes from repairi

ing summer 2order to test

Figure 7). Annems during th

ain reason foras damaged. ne during daytest. Both fu

ed total 26 00

ics in Construc

e. Cartesian cplex mechanise co-ordinatede surface, e.g. and excavato

lized control aides of the montroller AreIO modules.

nnected directl

h means that aules can workdevices, whic

ssor from Fr

module. ntrol and swirrent settings a

hine is equippepositioning

values are tranode (M modeing model.

2008 the systemt the whole c

nual average he survey or m

r repairing was

ytime. All thull 3D profilin0 m2 of repair

ction (ISARC 2

ontrol based m and when

d manual controad. Some e

or [Makkonen

and distributemachine, Figura Network). ADevices supp

ly to the CAN

all the controlk as a CAN mch increases rreescale. That

itches for chaare presented

ed with PC coand reads setnsferred to the), height of c

m has been inoncept from

daily traffic making the 3D

s the rutting. S

e ROADMOng and automring work.

009)

on solving invmanual contr

trol and automexamples are 2006].

ed IO. The core 6) and fouAnalogue sen

porting CANON bus. Use of C

l tasks are hanmaster and haeliability. Thet offers 60 M

anging operatin a graphical

omputer and t values from

he machine cocutter (or spre

n use for aboutthe surveying

in the test si profile due to

Some sections

OTO sensors matically contr

verse rol is matic road

ontrol ur IO nsors, Open CAN

ndled andle

e user MIPS

tional l user

RTK m the ontrol eader)

t two g and

ite is o low

s had

were rolled

4.2 Testing of thDuring th

ROADMOTOon the right border). Ultraroad border.

In figure 8right side of t2 m. In the upand measured

The closedvalue is usedcylinders is ab

Relative aaccuracy of thTemperature 0.18% / ºC. Tcase of heatedwell, but the bof the height,work is still n

4.3 Quality meaAfter the

Internatiol RoIRI was m

ROADMOTOjust doing resmoothed in t

5. Conclusio

The ROAsince June 20operations inthe design dathis concept. automatic ma

Although automatic conautomatic fea

the control systemhe tests diffeO machine wside of the masonic contro

8 the functionthe machine. Dpper graph sed value l2m of td-loop-contro

d [ u = Kp·e +bout ± 1 mm. accuracy of thhe height conchanges affecTo compensad asphalt air nbig temperatu, that is set du

needed.

asurements repairing wo

oughness Indemeasured befoO was doing a

epairing of ththe repairing w

ons

ADMOTO ma008. The repaincluding machata, a road repAll the phase

achine controlthere exists sontrol method atures accordi

Aut

m erent operatin

was driven alonmachine (centr

l is not applic

ning of the coDuring the 20t value for he

the position ofol is done over+ Kv·v ]. Samp

he height conntrol could notct to the outpate this integranear the surfacure difference uring calibrati

ork quality meex (IRI) [Sayerore and after automated rep

he pavement. work to be 0.8

achine equippiring process o

hine control apairing model,es from the sul were compleome problemis applicable. ing to the situ

tomation and Ro

3

Figure 7

ng modes of ng the left lanre of the roadcable if the su

ntrol system i0 second time ight h2 and mf the cylinder r CAN bus. Spling time Ts

trol is about t be verified p

put of the ultrated temperatce can be veryin the air incron of the syst

easurement wrs 1986]. the repairing

pairing. IRI leHigh origina

8 mm/m.

ped with the cof the road pnd quality co, in automaticurvey using a ted. Finally acs e.g. in the caIt is also impuation. This w

obot Application

322

. Test site.

the system ne of the roadd) and manuaurface of the

is shown wheperiod in the

easured value are shown.

Simple P-contrwas 20 ms.

±5 mm. Theproperly. Heigrasonic sensorture compensy hot (about 1reases noise oftem, changes.

was done. On

g job (Figure evel 0.8 is gooal value betw

control systemavement consntrol tasks. T

c machine concar mounted

chieved qualityalibration of tortant to that was also take

ns

were tested. d. In this caseal control was

road is very

n model conte figure the ma

h2m are shown

roller and feedAccuracy of

e drawback oght control is rs, because thation is used 50 ºC). Ultrasf the measurem To overcom

ne way to me

9). In the Figod and it is noeen the dista

m presented insists of survey

The developedntrol. Prototypsurveying systy was measurethe machine c

the user can en into accou

Typical situa ultrasonic co

s used on the uneven, whic

trol (M mode)achine movesn. In lower gr

d forward of the position

f the test wabased on ultr

he velocity of in ultrasonic sonic sensors ment. Also the this problem

asure quality

gure 9 is onet easy to go u

ances from 60

n this paper hying, design od control systpe test were atem to the reped. ontrol systemchoose betwent during the

ation was thontrol was use

left side (roach is typical fo

) is used on th forward abou

raph set value

the velocity scontrol of th

s that absolutrasonic sensorsound changesensors. In thstill work quit

he zero positiom developmen

of the road

section wherunder that valu00 to 700 wa

has been in usf repairing, sitem can explo

arranged to tepair design an

m, the presenteeen manual ane design of th

at ed ad or

he ut l2

et he

te rs. es he te

on nt

is

re ue as

se te

oit st

nd

ed nd he

control sybeforehan

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

41

IRI (

mm

/m)

26th Internatio

ystem and the nd.

10 430 450 470

onal Symposium

presented sys

Figure 8. Au

Figure 9. Ro

0 490 510 530

B

m on Automati

stem is also us

utomatic cont

oad quality be

0 550 570 590

D

Before Repair

ion and Roboti

323

seful in work

trol of the righ

efore and after

610 630 650

Distance (m)

ring A

ics in Construc

sites, where n

ht side of the

r the repairing

670 690 710

After Repairin

ction (ISARC 2

no surveying a

cutter.

g work.

730 750 770 7

ng

009)

and design is m

790

made

Automation and Robot Applications

324

During the short period of testing it was not possible to fully evaluate the benefits of the system. The next goal is to take the presented process into common use.

References [1] Kilpeläinen, P., Nevala, K. (1999). “Design and simulation of a road scraper control using the

TeleGripTM simulation program”, Proceedings of Scandinavian Symposium on Robotics 1999. Oulu, 14 - 15 Oct. 1999. pp. 232 - 239.

[2] Kilpeläinen, P. Nevala, K. Heikkilä, R. (2004). ”Development of a control system for road construction automation applications”, Proceedings of 21st International Symposium on Automation and Robotics in Construction. Jeju, 21 - 25 Sep. 2004. IAARC, pp. 65 – 70.

[3] Makkonen, T., Nevala, K., Heikkilä, R. (2006). ”3D model based control of an excavator”. Automation in Construction, vol. 15, 5, pp. 571 – 577.

[4] Sayers, M., Gillespie, T., Paterson, W. (1986). “Guidelines for Conducting and Calibrating Road Roughness measurements”. World Bank Technical Paper, vol. 46, 5, ss. 1 - 12