Development of a Control System for a Multipurpose Road Repairing Machine

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

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:

[email protected]

2Destia Ltd, P.O.Box 444,Oulu, FIN-90101,Finland;email: [email protected]

3Destia Ltd, Turuntie 207,Espoo, FIN-02740,Finland;email: [email protected]

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 the new asphalt la asphalt is used The idea o the worst def the road geom 100% and ov repairing mod

ROADMO remixer mach The spreader

2. Surveying Surveying and applying

2.1 Survey and Surveying system contai

e crushed old ayer on top of d on the botto of the machin fects of the ro metry is neede verall cost savi

del.

OTO machin hine. In this ca

in the middle

Figu and Modelin and modellin automation fo

d 3D model of the road is ins laser, full i

pavement. A f the old layer om layer.

ne is to use ol oad and to mi ed about 20 kg

ing is about 1 ne is also equ ase old crushe e of the machi

re 1. Multipur ng

ng phases (Fig or the repairin

Figure 2.

s done by spe inertial and G

Asphalt spread r. So two layer

ld heated and inimize the us g/m². It is est 5%. In this ca uipped with a ed pavement i ne is not used

rpose road rep

gure 2) are im ng of the pave

Process to su

ecial measurem GPS devices. S

er in the back rs of asphalt a d crushed pave

se of new pav timated that w ase cut and fil a mixing drum

s mixed with d in the remixe

pairing machin

mportant when ment.

urvey and repa

ment system m System collect

k of the machi are done at the

ement as muc vement. Typic with automatio

ll operations a m and it can the new asph er work.

ne (ROADMO

n designing re

air the road.

mounted in a ts all the data

ine is used fo e same time a ch as possible ally new mass on this can be are done acco

be used also alt mass in the

OTO).

epairing mode

car (Figure 3) from the sen

r spreading th and old crushe e for correctin s for correctin reduced 50%

rding to a roa o as an aspha e mixing drum

el for road sit

). Measuremen sors to the raw

he ed ng ng

% - ad

m. alt

te

nt w

data file d longitudin

Accura horizontal application thickness i 2.2 Design f Design specified.

done and and decide damages n The am possible to used for fi

When ROADMO Input data whole con coordinate 3. Machin 3.1 Main fu The de operation - N sites.

- Su machi The m vary on a possible.

The au middle of back of th Both th automatio

26th Internatio during the m nal road profile

acy for the 3D l direction gr ns. Ground p is needed.

for repair n for repair is

In the designi what the targe ed how to fix need special ac mount of asp o mill more th illing. GPR da

designing is OTO. Repairi a file defines t nstruction site e is given and ne Control Sy functions

evelopment w of the machin No surveying a urveying and ine control. Su machine contro

work site con utomatic cont f the machine.

he machine is c he cutter and on system con

onal Symposium measurement.

es are combin

D model in ve rid resolution penetrating ra

s done with ing phase of t et of repair is.

them. There ctions when re phalt used is

han needed fo ata is useful in

ready the p ing file is cre the milling dep e. If spreader it is related to ystem work of the co

ne. Two main and modelling modelling is urveying and m ol system shou nsiderably, cho trol system is . These are re controlled ma the spreader ntrols are cutt

m on Automati Next phase i ned in to one 3

Figure 3.

ertical directio n is possible adar can be u

terrain model the repair proc Different kin can be e.g. ed epaired.

optimized in or repairing an n optimization

plan is inser eated and RO pth and filling

is used in th o milled surfac

ontrol system situations are g is done befo s done before modelling are uld be applicab oosing betwee

used for con eferenced belo anually and is n

are moved by ting depth (or

ion and Roboti is the proces 3D surface of

Surveying sys on is provided e to choose used, if infor

lling applicati cess it has to b nds of things o dge drops, pot the designin nd there will b n.

rted in to th OADMOTO’s

g height for ea he repairing w

ce.

started with e:

orehand. This ehand and a

typically done ble for both o en manual and ntrolling the m

ow as the cutt not connected y two hydraul r height) h1 an

ics in Construc sing of the c f the road.

tem.

d by high reso to be high rmation about

ion. Design is be known wha of the paveme tholes or deep ng phase. E.g.

be sections wh he automated

control softw ach grid coord work, the spre

analysing diff s is typical situ

repairing mo e only in main of these situati d automatic op milling drum a ter and the sp d to the autom lic cylinders (F nd h2 and slop

ction (ISARC 2 collected data

olution laser u enough for t road structu

s done with at kind of rep ent condition p rutting. Var . there are se here the crush d pavement r ware can read dinate. The inp eader height l

ferent working uation in sma odel is availa n roads and lar ions. Because peration shou and the aspha preader. Asph mation system Figure 4). The pe θ of the cu

009)

a. Transversal

units (0.1 mm) terrain mode ure e.g. pave

the accuracy airing work w has to be surv rious kinds of ections where

hed asphalt w repairing mac d it as input put data file co level for each

g situations du all roads and w able for autom rge work sites

the circumsta uld be as flexib alt spreader in halt spreader in

.

e variables tha utter (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).

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 h^•₁ , h^•₂and

θ

In this kinematics demanding control ba grader [Ki 3.3 Hardw The co system co modules, such as ul protocol ( bus reduce

Contro by the use control ta interface performan

User in modes and interface.

3.4 Model b In the (Real Tim repairing m system via are contro 4. Prototy The sy months. I modelling 4.1 Test sit

Test si approxima traffic.

Condit edge drop Repairi mounted repair wor

26th Internatio case the mec s gives more g and high ac ased on real t ilpeläinen 199 ware and software

ontrol system onsists of two which are co ltrasonic senso (e.g. directiona

es the amount ol software is er interface m sks. The ben module is ba nce for contro

nterface mod d set values (e

based control mo model based me Kinematic) model accord a CAN bus. If olled automatic

ype Test ystem was take In October 20 to the autom te

ite is typical ately 2000 veh tion of the tes s, but only 20 ing work wit and raw data rk were succes

onal Symposium chanism that i

advances wh ccuracy is nee time positioni 9], stabilizatio e

architecture i o user interfac onnected toge

ors, are conne al valves and i

t of wiring con implemented module. Both nefit of this is

ased on a 16 ol tasks.

dule consists o e.g. height can

ode

control mode ) GPS receiv ding to the po f the user has cally accordin

en in use in Ju 008 a test wa matic machine c

road in sout hicles. There w st site was goo 0 – 30 cm from

th ROADMO a was collecte

ssful. The test

m on Automati is controlled i en controlling ded. Some ap ing and 3D m on cutter [Kilp

is a combinati ce modules (o ther using a ected to the C inclination sen

nsiderably.

in a centraliz of the user in

the duplicatio 6 bit digital

Figure 6. Us of joystick fo n be set in 1 m

e the ROADM ver. PC comp osition of mac s selected mod ng to the value

une 2008. Duri s arranged in control.

th Finland (F were no proble od and the ma m the edge wa OTO was don

d during the t site considere

ion and Roboti is quite simple g more comp pplications are models of the peläinen 2004]

ion of central one in both s CAN bus (C CAN bus via nsors) are con ed way, which nterface modu on of control

signal proces

ser interface m or manual con mm steps). Cur

MOTO mach puter handles chine. These v

del control m es from repairi

ing summer 2 order to test

Figure 7). Ann ems during th ain reason for as damaged.

ne during day test. Both fu ed total 26 00

ics in Construc e. Cartesian c plex mechanis e co-ordinated e surface, e.g.

and excavato

lized control a ides of the m ontroller Are IO modules.

nnected directl h means that a

ules can work devices, whic ssor from Fr

module.

ntrol and swi rrent settings a

hine is equippe positioning values are tran

ode (M mode ing model.

2008 the system t the whole c

nual average he survey or m r repairing was ytime. All th ull 3D profilin

0 m² of repair

ction (ISARC 2 ontrol based m and when d manual cont road. Some e or [Makkonen

and distribute machine, Figur

a Network). A Devices supp ly to the CAN all the control k as a CAN m ch increases r reescale. That

itches for cha are presented

ed with PC co and reads set nsferred to th e), height of c

m has been in oncept from

daily traffic making the 3D s the rutting. S

e ROADMO ng and autom ring work.

009)

on solving inv manual contr trol and autom examples are

2006].

ed IO. The co re 6) and fou Analogue sen porting CANO N bus. Use of C l tasks are han master and ha eliability. The t offers 60 M

anging operat in a graphical

omputer and t values from he machine co

cutter (or spre

n use for about the 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

m the RTK ontrol eader)

t two g and

ite is o low s had were rolled

4.2 Testing of th During th ROADMOTO on the right border). Ultra road border.

In figure 8 right side of t 2 m. In the up and measured

The closed value is used cylinders is ab

Relative a accuracy of th Temperature 0.18% / ºC. T case of heated well, but the b of the height, work is still n 4.3 Quality mea

After the Internatiol Ro IRI was m ROADMOTO just doing re smoothed in t 5. Conclusio

The ROA since June 20 operations in the design da this concept.

automatic ma Although automatic con automatic fea

the control system he tests diffe

O machine w side of the m asonic contro 8 the function the machine. D pper graph se d value l2m of t d-loop-contro d [ u = Kp·e + bout ± 1 mm.

accuracy of th he height con changes affec To compensa d asphalt air n big temperatu , that is set du needed.

asurements repairing wo oughness Inde measured befo O was doing a epairing of th

the repairing w ons

ADMOTO ma 008. The repai ncluding mach ata, a road rep All the phase achine control there exists so ntrol method atures accordi m

erent operatin was driven alon machine (centr

l is not applic ning of the co

During the 20 t value for he the position of ol is done over + Kv·v ]. Samp

he height con ntrol could not

ct to the outp ate this integra near the surfac ure difference

uring calibrati

ork quality me ex (IRI) [Sayer ore and after automated rep he pavement.

work to be 0.8

achine equipp iring process o hine control a pairing model, es from the su l were comple

ome problem is applicable.

ing to the situ

Figure 7

ng modes of ng the left lan re of the road cable if the su ntrol system i 0 second time

ight h2 and m f the cylinder r CAN bus. S pling time Ts

trol is about t be verified p put of the ultr ated temperat ce can be very in the air incr on of the syst

easurement w rs 1986].

the repairing pairing. IRI le High origina 8 mm/m.

ped with the c of the road p nd quality co , in automatic urvey using a ted. Finally ac s e.g. in the ca

It is also imp uation. This w

. Test site.

the system ne of the road d) and manua urface of the is shown whe

period in the easured value are shown.

Simple P-contr was 20 ms.

±5 mm. The properly. Heig rasonic sensor ture compens y hot (about 1 reases noise of tem, changes.

was done. On g job (Figure evel 0.8 is goo al value betw

control system avement cons ntrol tasks. T c machine con car mounted chieved quality

alibration of t ortant to that was also take

were tested.

d. In this case al control was road is very n model cont e figure the ma

h2m are shown roller and feed Accuracy of e drawback o ght control is rs, because th ation is used 50 ºC). Ultras f the measurem

To overcom

ne way to me 9). In the Fig od and it is no

een the dista

m presented in sists of survey The developed ntrol. Prototyp surveying syst y was measure the machine c

the user can en into accou

Typical situa ultrasonic co s used on the uneven, whic trol (M mode) achine moves n. In lower gr d forward of the position f the test wa based on ultr he velocity of in ultrasonic sonic sensors ment. Also th e this problem

asure quality gure 9 is one

t easy to go u ances from 60

n this paper h ying, design o d control syst pe test were a tem to the rep ed. ontrol system

choose betwe nt during the

ation was th ontrol was use left side (roa ch is typical fo

) is used on th forward abou raph set value

the velocity s control of th s that absolut rasonic sensor sound change sensors. In th still work quit he zero positio m developmen

of the road section wher under that valu 00 to 700 wa

has been in us f repairing, sit em can explo arranged to te

pair design an m, the presente een manual an e design of th at ed ad or he ut l2

et he rs. te es he te on nt

is re ue as

se te oit st nd ed nd he

control sy beforehan

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

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

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