Autonomous Weeders for Christmas Tree Plantations

8/2/2019 Autonomous Weeders for Christmas Tree Plantations

1/86

Pest icides Resear ch Nr. 59 2002Bekmpel sesmiddel f or skning f r a Mil jst yr el sen

Autonomous weeders for Christmas

tree plantations - a feasibility study

Henrik Have (Editor)

With contributions from:

Simon Blackmore1Henrik Have1, Bent Keller2, Spyros Fountas1,Henning Nielsen1and Frans Theilby21) The Royal Veterinary and Agricultural University, Section ofAgroTechnology2) The Danish Forest and Landscape Institute, Department ofForestry


2/86

T he D anish Environmental Protection Agency will, when opportunity

offers, publish reports and contributions relating to environmental

research and development projects financed via the Danish EPA.

Please note that publication does not signify that the contents of the

repor ts necessarily reflect the views of the Danish EPA.

T he repor ts are, however, published because the D anish EPA finds that

the studies represent a valuable contribution to the debate on

environmental policy in D enmark.


3/86

3

Contents

PREFACE 5SAMMENFAT NIN G 6SUMMARY 81 INT RODUCT ION 102 THE ECONOMIC S IGNIFICANCE OF CHRISTMAS TREEPRODUCT ION 12

2.1 STRUCTURE 122.2 EXPORT 122.3 ECONOMICS 12

3 CURRENT CULT IVAT ION MET HODS 133.1 ESTABLISHMENT 133.2 WEED CONT ROL METHODS 14

3.2.1 Chemical 143.2.2 M echanical 14

3.2.2.1 Conditions for mechanical weeding 163.2.2.2 T ools for mechanical weeding 17

3.2.3 Animals 19

3.3 HARVESTING 193.4 CONT RACT ORS AND CON TRACTOR COSTS 193.5 COSTS OF THE VARIOUS OPERATIONS 19

4 WEED CONTROL REQUIREMENTS IN CHRISTMAS TREEPRODUCT ION 214.1 METHOD 214.2 RESULTS 22

4.2.1 T he average height of the trees. 224.2.2 Root collar diameter of the trees 234.2.3 Health score of the living trees. 234.2.4 Soil moisture 24

4.2.4.1 T nballegrd 244.2.4.2 Wedellsborg 25

4.3 OT HER INVESTIGAT IONS 264.4 D ISCUSSION 264.5 SUB CONCLUSION 27

5 SPECIFICAT ION OF ST AKEHOLD ER REQUIREMEN T S 285.1 T HE CURRENT LEVEL OF SATISFACTION WITH SPRAYING ANDMECHANICAL WEEDING T ECHNOLOG Y 285.2 ST AKEHOLDER REQUIREMENTS T O AN ACW 295.3 SUB CONCLUSION 30

6 SPECIFICATION OF TECHNICAL REQUIREMENTS FOR ANAUT ONOMOUS CHRIST MAS T REE WEEDER 316.1 OVERVIEW OF CULT IVATION M ETHOD OLOGY 31


4/86

4

6.2 D EFINITION OF OPERATION SCENARIOS FOR T HE ACWDEVELOPMENT 326.3 D EFINITION O F WORK T ASKS 326.4 SUB CONC LUSION 33

7 AUTONOMOUS VEHICLE TECHNOLOGY A LITERAT UREREVIEW 347.1 AUT OMATIC AND AUT ONOM OUS VEHICLES IN AGRICULT URE,FOREST RY AND HORT ICULTURE 357.2 EVOLUT ION OF AUTOM ATIC STEERING FOR AGRICULT URALVEHICLE 367.3 SENSORS FOR N AVIGATION 37

7.3.1 Real time Kinematic global position system 37 7.3.2 Computer vision 37 7.3.3 Supplementary sensors 387.3.4 T he leader-cable principle 387.3.5 Laser-based principles 387.3.6 Other principles 39

7.4 SAFETY

397.5 SYSTEMS ARCHITECTURE 407.5.1 T hree-layer architecture 417.5.2 T he Saphira architecture 41

7.5.3 T he animate agent architecture 427.5.4 Behavioural reactive system architecture 42

7.6 SUB CONCLUSION 42

8 PROPOSED OUTLINE OF AN AUTONOMOUS CHRISTMAST REE WEEDER 448.1 WEEDER T OOL 448.2 OPT IMUM OPERAT ION PATT ERN AND VEHICLE POSIT IONING 44

8.3 VEHICLE CONCEPTS 458.3.1 T he beetle type 458.3.2 T he portal type 47

8.4 VEHICLE OPERAT ION PROCEDURE AND T RANSPORT 478.5 SYSTEM ARCHITECT URE 488.6 SUB CONCLUSION 48

9 ESTIMATED PERFORMANCE OF SELECTED ROBOTCON CEPT S 499.1 ENVIRONMENT AL BENEFITS 499.2 EFFECTIVENESS OF WEED C ONT ROL 499.3 MACHINE CAUSED LO SSES 49

9.4 SAFETY 509.5 ESTIMAT ED AREA CAPACIT Y 509.6 ESTIMAT ED COSTS 519.7 ADD ED VALUES 549.10 SUB CONCLUSIONS 54

10 CONCLUSION S 5511 REFERENCES 57Bilag A: A specification of behavioural requirements for an autonomoustractorBilag B: Prop osed Systems architectureBilag C: General motivation of using autonomous vehicle systems.


5/86

5

Preface

Agriculture, horticulture and forestry have in the past benefited from asuccession of technological developments that have brought greaterproductivity and econom ic efficiency. Historically, the emphasis of thesedevelopments has been on the mechanisation to increase work rates throughuse of larger and more powerful machines. The newer information basedtechnologies have already been used for some time to improve functions andcontrols of machinery, especially for spatial graduation of treatments. Thesetechnologies have now reached a stage which seems to make autonomou s fieldmachinery and individual treatment of plants realistic.

Mechanical weeding in C hristmas tree plantations is a well suited area tobegin development of such equipm ent because of the relatively large size of

the plants, the difficulties of using standard agricultural machinery andpresent extensive use of herbicides.

T his report presents results from a study on the feasibility of developing anautonomous Christmas tree weeder, including the technical, economic andenvironmental aspects, as well as the possibilities of using it for automatic datacollection for management decisions.

T he study was carried out in a collaborative project between T he RoyalVeterinary and Agricultural University, Section for AgroTechnology and TheResearch Centre of Forestry and Landscape, Department of ForestManagement. The study has been financially financed by the Ministry ofEnergy and Environment.

I acknowledge with thanks the valuable assistance we have received from:T omas Nordfjell, T he Research Centre of Forestry and Landscape,Hans-Werner G riepentrog, The Royal Veterinary and Agricultural UniversityBent K. C hristensen, T he Christmas T ree Growers Association,Sten Gellerstedt, Forestry Robot Research Group, Agricultural UniversitySweden,Andreas Bergstedt, The Royal Veterinary and Agricultural University,Mogens Blanke, T he T echnical University of DenmarkSren H onor, Danish Forest and N ature Agency,Esben Pedersen, Silvatec A/S, andMorten Srensen, Wedellsborg Estate.Dvoralai Wulfsohn has checked the language.

Henrik Have


6/86

6

Sammenfatning

Den danske juletrsproduktionen dkker ca. 31000 ha og har en samletomstning p 500-600 mill. kr. Omkring 4100 producenter er involveret. D efleste har mindre end 1 ha.

For at opn god vkst og kvalitet af juletrer er det ndvendigt med hyppigukrudtsbekmpelse. Dette gennemfres i langt de fleste tilflde medherbicider, hvilket frer t il helt rene kulturer med ringe levemuligheder forandre arter. Endvidere medfrer det risiko for forurening af grundvandet.

For at reducere disse problemer er der udviklet maskiner til mekaniskrenholdelse. De frste ca. 2 r efter plantningen kan anvendes skaldtelangfingerharver, der kan bekmpe mindre ukrudt uden at beskadige trerne

for meget. I de efterflgende vkststadier er det ndvendigt at anvenderadrensere eller andre t ilsvarende maskiner. D isse maskiner er imidlertid fordet meste kun i stand til at bearbejder jorden m ellem rkkerne. Bekmpelsener sledes mindre effektiv. Samtidig er omkostningerne omkring dobbelt shje som for sprjtning. Endvidere har jordbearbejdningen uhensigtsmssigevirkninger i form af udvaskningen af nringsstoffer og get jorderosion.

Nrvrende projekt prsentere muligheder og ideer for udvikling af en lilleautonom m askine til mekaniske ukrudtsbekmpelsen p en mere miljvenligmde. Studiet er indledt med en undersgelse af det minimale behov forukrudtsbekmpelse p basis af lbende forsg ved Forskningscenter for Skovog Landskab. Den viste, at det er tilstrkkeligt at bekmpe ukrudtet i cirkleraf 40 cm radius omkring det enkelte tr, hvilket svarer til ca. 40% af det totaleareal. D et ikke bekmpede ukrudt p den vrige del af arealet m gerne blivestende, da det giver en vis l og derfor har en positiv virkning p vksten aftrerne.

P basis af et litteraturstudium vedrrende robotteknologi er der i projektetudarbejdet koncep tforslag til nogle forskellige systemer, herunderoperationsmnstre, design af kretj, behov for sensorsystemer, databaser ognavigationssystem. Ogs et estimat for omkostningerne ved brug af en sdanmaskine er blevet udarbejdet for at kunne sammenligne medmaskinstationstakster for nuvrende operationer.

Undersgelsen konkluderer, at det ud fra et tekniske synspunkt er realistisk atudvikle en sdan lugemaskine, og at den vil kunne op fylde mlstningen omat renholde arealet omkring trerne og at lade det meste af floraen st.Endvidere konkluderes det at maskinen for sm omkostninger vil kunneudvikles til i forbindelse med arbejdet, at indsamle data om det enkelte trvedrrende tilstand og kvalitet med henblik p anvendelse i driftsledelsensbeslutningstagning og dokumentation.

Det vurderes, at maskinen vil have vsentlige lavere miljmssige effekter oggive en lidt bedre tilvkst end de nuvrende teknologier (sprjtning ogmekanisk ukrudtsbekmpelse). Det vurderes endvidere, at maskinkonceptet

vil have betydelig interesse i forbindelse med ukrudtsbekmpelse i andrespecialkulturer, som f.eks. frugt- og brplantager.


7/86

7

Det valgte koncept bygger p njagtig bestemm else af positionerne for deenkelte trer og for den autonome maskine ved hjlp af et meget prcistglobal positioneringssystem ( RT K G PS) . T rernes position tnkes mlt ogkortlagt au tomatisk i forbindelse med plantn ingen, idet et elektronisk systemmler positionen, hvor de placeres i jorden. Det fremkomne kort anvendesderp gennem hele plantagens levetid af den autonome maskinen somgrundlag for navigation under arbejdet. Ud over positionen vil der ogs vrebehov for lbende mling af en rkke andre parametre for tilstanden omkringog p maskinen t il brug for navigation, styring og kontrol, herunder etintegreret sikkerhedssystem, der skal forhindre sammenstd medforhindringer, beskadigelse af trer og materiel, samt tilskadekomst afmennesker og dyr. Disse styrings - og kontrolfunktioner er struktureret i ensrlig systemarkitektur.

Til bekmpelse af ukrudtet er valgt en rotorklipper, der har mindreenergiforbrug og nsten lige s god effekt som et jordbearbejdningsvrktj.Som operationsmnster er valgt en tilnrm elsesvis retlinet bevgelse langs

den enkelt rkke, hvilket har vist sig at vre det mest effektive og konomiske.Det er endvidere valgt, at styre rotorklipperens horisontale finpositionering iforhold til det enkelte tr via en mekanisk eller optisk sensor.

T o typer af platforme er undersgt: 1) En lille, lav 4 hjulet maskine, der kankre under grenene og ind mellem trerne, og 2) En portal maskine, der kanskrve over en rkke og bearbejde p begge sider. Den frstnvnte typevurderes at vre den enkleste og mest velegnede.

Det er valgt at opbygge systemarkitekturen i flere lag, der bl.a. omfatterkoordinatoren, som er det menneske, der stter maskinen i gang og lbendeer i kontakt med m askinen via en PC, vejlederen, der er en computer p

maskinen, som primrt via mode changer srger for at maskinen opfrersig hensigtsmssigt (navigation, rutep lanlgning, sikkerhedscheck ogudfrelse af arbejdsopgaver mm.) , sekundrt vejleder de forskellige stadier afmaskinens sikkerhedsadfrd og endelig kommunikerer maskinens situationeller tilstand til koordinatoren.

Med hensyn til omkostningerne ved anvendelsen af maskinen m der i defrste tid efter udviklingen forventes et noget hjere niveau end for de kendteteknologier. M en den generelle udvikling i elektroniske komponenter ogudstyr, sammen med stigende produktionstal m forventes ret hurtigt atbringe omkostninger ned p et niveau mellem de nuvrende mekaniske ogkemiske metoder. Ud over dette vil maskinen antagelig kunne givetillgsvrdier i form af indsamlede data for det enkelte tr som grundlag fortil driftslederens beslutninger og kundeinformation.


8/86

8

Summary

T he Danish Christmas trees production covers about 31000 ha and has anannual turnover about 500-600 million D KK . About 4100 producers own thisarea. T he majority have less than 1 ha.

To obtain good growth and quality frequent weed control is essential. This ismainly done by application of chemicals, which leads to clean stands withpoor living conditions for other species and r isks of pesticide leaching.

To reduce these problems of chemical weed control machinery has beendeveloped for mechanical weeding. The first approximately two years afterplanting it is possible to use special spring tooth harrows with long flexibletines, which are effective in control of small weed p lants without damaging the

trees too much. In the following growth phases it is necessary to use row cropweeders. T hese weeders, however, are generally only able to control weedsbetween the rows an also less effective and more costly to use than herbicides.In addition the tillage process may cause negative effects of nutrient leachingand soil erosion.

This project report presents the feasibility of developing a small autonomousmachine for mechanical weed control in a more environmentally friendly way.T he study was initiated by defining the minimum requirements to weedingusing results from an ongoing investigation at The Danish Forestry andLandscape Research Institute. T his investigation has shown, that maximumtree growth and developm ent is achieved by weeding of concentric circle areasof only 40 cm radius around each tree. The left over weeds on the remainingarea have beneficial effects on the trees because of shelter effects.

On the basis of a literature review the idea and overall concepts of a smallautonomous weeding machines is outlined, including suggestions of operationpatterns, physical designs of the vehicle and the system architecture withsensors, databases, navigation system, and actuators. Also the likely costs ofusing such a m achine were estimated and compared to currents contractorrates.

The study has led to the conclusion, that it from a technical point of view is

realistic to develop an autonomous Christmas tree weeder and possible tofulfil the aim of cleaning only part of the area. In add ition the mach ine isconsidered suitable for collection of tree specific data concerning conditionand development for use in p lantation management and productdocumentation.

It is estimated that the machine will have significantly lower environmentaleffects and facilitate better tree growth than present technologies (sprayingand mechanical weeding). Further on it is estimated, that the machine asconcept will have considerable interest for weed control in other specialcultures as fruit and berry plantations.

T he chosen concept is based on centimetre precision determination of theposition of each tree as well as the current position of the ACW by means of a


9/86

9

RTH GPS (Real Time Cinematic Global Positioning System). The positionsof the trees are to be measures automatically during machine planting byrecording of the position where the tree is placed in the soil. The mapproduced in this way can then be used by the autonomous machine as basisfor navigation during weeding over the entire lifetime of the plantation. Apartfrom the position it would also be necessary to measure a number of otherparameters of the conditions around and in the machine itself for navigation,steering and controls, including an integrated safety system, which is to avoidcollisions with obstructions, and injury of humans, animals and trees. Thesesteering and control functions are structurized in the system architecture.

As weed control tool is chosen a rotor cutter similar to those used in certainlawn mowers, as this needs less energy than tillage, and the effect is nearly thesame as for mechan ical weeding. As operation pattern is chosen a nearly linearmovement along the rows as this was found to be the most economic. It wassuggested to control the horizontal fine position of the rotor relative to thesingle tree by means of an optical or mechanical distance sensor.

Two platforms were considered: 1) A small, low machine being able to movebelow the branches and in between the trees. 2) A portal machine that have aset of wheels and a working tool on either side of the row. T he first isestimated to be the simplest and best suited.

T he system architecture, which is to be built up in several layers, comprises acoordinator, who is the person that runs the vehicle via a PC, a supervisor,which is a computer, that primarily via a mode changer controls themachine to behave appropriate, i.e. to navigate, to do route planning, safetycheck, and secondary supervision at the different stages of the safetybehaviour, as well as communicating the situation and conditions to thecoordinator.

As regards economy it is estimated that the continued price reductions ofelectronic equipment together with increasing production numbers ratherquickly will bring the costs dawn at a level between the present costs ofmechanical weeding and spraying. On top of that added values may beachieved in terms of tree specific data for management decisions andcostumer information.


10/86

10

1Introduction

T o achieve good growth and quality of Christmas trees it is traditionallyconsidered necessary to weed very intensively to compensate for the weakcompetitiveness of Christmas trees, especially Nordmanns fir, which is themost grown species. Intensive weed control is in m ost cases achieved byapplication of persistent soil herbicides with a broad spectrum of effect.

In resent years many growers have shifted to screened spraying with foliageacting herbicides. T he rate of application is usually close to the allowedmaximum, and the spraying takes place every year over the entire productionperiod (app. 10-12 year).

T he extensive use of herbicides is alarming in itself, especially seen in the light

of the latest investigations, which show that it is leaching into the soil(Jacobsen et al. 2000). Leaching of nutrients especially nitrates from a soilalmost without vegetation is also considerable, as the sparsely placed trees areonly able to utilise an insignificant fraction (Rubow et al. 2000). Thebiodiversity of total cleaned areas is probably very low as a result of the lack offood and hiding places.

This makes Christmas tree plantations a high priority area for alternativeweeding methods (Bichel, 1998). Some growers have moved in that directionduring the past years and are beginning to use mechanical weeding (Keller,1997) . T he weeding imp lements developed for the purp ose are well able toremove weeds between the rows, but mostly not in the rows, where the need isgreatest. Most of the implements are rather heavy, of low capacity, and costlyto use. T he general strategy of weeding using these machines is the same asfor spraying: to achieve a total clean area. T herefore, this also leads toproblems of leaching and relatively low biodiversity.

T he pu rpose of the p resent project was to investigate the feasibility ofdeveloping a light, autonomous weeding machine being able to performmechan ical weed control more com petitively within a relatively short period oftime and also to reduce the environmental problems associated with thepresent methods. T he machine should have an acceptable behaviour, be ableto operate unattended and safely for longer periods of time.

T he project comprises: a description of the present Christmas tree cultivation systems, an investigation of the need for weed control around single Christmas

trees, specification of the conditions in which the autonomous systems should

work, specification of the stakeholder requirements of an autonomous system, specification of the technical requirements of the system two proposed machine concepts, a proposal for a system architecture, including navigation system, safety

system and overall control system, an evaluation of the proposed concepts compared to present methods,including biological, technical, economic, environm ental and safety issues


11/86

11

as well as the possibility to achieve added values by collection ofinformation on the trees for management decisions.


12/86

12

2The economic significance ofChristmas tree production

Bent K ellerThe Danish Forest and Landscape Institute, Department of Forestry

According to the registrations at T he Production Fee for Christmas T rees andGreenery (PAF) that the production area in D enmark is 31.370 hectare.From this Nordm anns fir constitutes the major part ( approx. 22.000 hectare)and noble fir approximately 9.000 hectare. Further there are some specialproductions of cypress, Serbian spruce, holly, cryptermeria and others. Andfinally there is Norway spruce, which is either produced in field p lantations or

taken from forest plantations as thinning trees. The total area is thus app.35.000 hectare.

2.1 St r uct ur e

About 4.100 producers are registered, of which the majority are operatingwith area units < 1 hectare. It is estimated that the size of an averageproduction unit is about 1,5 hectare; however, there has been a tendencytowards larger production units during the later years.

2.2 Expor t

Denmark is the country in Europe that exports the most Christmas trees andgreenery cuttings. Each year 6-8 million Christmas trees and 30 35.000tonnes of greenery cuttings are exported. T he major buyer is Germany with50 60% of the Christmas trees and 70 75% of the greenery cuttings. Nextin place are France, Austria, England, Switzerland and Norway.

2.3 Economics

T he annual turnover is 500-600 million D KK and m akes up 50% of the totalturnover of Dan ish forestry. For each producer this involves a very intensiveproduction in which about 100.000 D KK per hectare is tied up in one cropfor a period of 10 years. An essential condition for a reasonable yield of theinvestment is that the plantation is weeded. Today this is to a great extent(app. 70% of the area) done chemically, amounting to 20-25% of the totalcultivation costs equivalent to around 25.000 D KK per hectare.


13/86

13

3Current cultivation methods


T his chapter provide an introduction to the most common methods ofChristmas tree production in D enmark with focus on mechanical weeding.Other control methods are described briefly.

3.1 Establ ish ment

At present Christmas tree plantations are established on considerable areas.T he majority of the plantations are established on previously cultivated land,

and only a small part on forestland. When establishing plantations the area isin most cases prepared with thorough brushwood clearing and stumping,sometimes followed with several haulages with a spring tine harrow to removeirregularities. The area is on the whole ready to be treated like previouslycultivated land.

T oday three cultivation methods are used:

1. Clear cutt ing (p lanting - harvesting of all trees planting) . T his isthe most prevalent cultivation system. It is highly systematic and iseasy to manage regarding planning. The planting is generally donewith planters. The planting distance is typically 1,2 x 1,2 m, equal to

app. 6.000 p lants per hectare (F igure 3.1).

2. Current planting (planting every time a tree is cut). Planning withinthis cultivation system is much more difficult because the trees in eacharea are of different ages and thus have different demands regardingtreatment. H owever this system gives a more even d ivision of expensesand income. Here planting is always carried out manually. Themethod has some b iological as well as environmental advantages.

3. Regeneration ( new trees develop from the stumps of the cut trees).This method is only used to some extent because it is technically

difficult, hard to manage, and often gives a poorer quality. Theadvantages are that there are no planting expenses, and as the plantalready is established there will be no period of stagnation ofestablishment.


14/86

14

Figur e 3.1. Nor mann s f ir s pl ant ed in r ows.

3.2 Weed cont r ol met hods

T here are advantages as well as disadvantages to the presence of weeds in thecultivation area.

Advantages: Leaching of nutrients is diminished Microclimate is improved Risk of wind- and water erosion is diminished T he biodiversity is improved

Disadvantages: Increased competition for water, light and nutrients Physical damage such as wear on the trees G rass vegetation increases the risk of spring night frost damage Comp licates trafficking

Normally the disadvantages are rated to carry great weight and this is why avery intensive weeding is started. T his weeding can be chemical, mechanicalor using animals.

3.2.1 ChemicalChemical weeding is by far the most used weeding method. Traditionallyslowly degradable broad-spectrum soil weed control was used. The rate ofapplication was always close to or the m aximum allowed dose, and theapplication was repeated each year throughout the rotation, which was 10-12years. Several of the traditionally used persistent soil weed controls are noweither forbidden or are being re-evaluated, and many of the producers havenow started using screened spraying with leaf spray. The purpose of weedingis still the same to keep the cultivation totally or almost free weeds.

3.2.2 MechanicalMechanical weeding of Christmas trees is the method that at present is


15/86

15

regarded to be the best alternative to chemical weeding. Mechanical weedingin plantations on previous forestland demands careful brush wood clearingand either stump removal or stumping, as the available tools have beenproduced for cultivated land without obstacles in the soil.

Mechanical weed control can be done by mowing plants above the ground orby soil tillage. Soil tillage is superior and the safest method. M owing does notreduce the risk of spring night frost like soil tillage, and the com petition fromthe weeds for water will not be completely reduced. A high occurrence ofnatural vegetation has a beneficial effect however, during winter as it reducesthe risk of winter frost damage, and a certain competition from the vegetationcan p robably have a beneficial restrictive effect on leader shoot growth .Therefore mowing can sometimes be justified, for example combined withother weed control methods ( sheep grazing) and at certain t imes in therotation or growth season. T he following descriptions focus on soil tillagemethods only.

T he weed controlling effect of soil tillage consists of detaching and covering of

the weed. T he effect of this depends on a number of factors: among othersthe weather, the soil structure and moisture, the structure of the weed and thestage of development of the weed, travelling speed and depth of tillage. Theeffect is very dependent on the weather and can be considerably reduced inmoist weather and soil. T he effect is highest when the weed control is carriedout in a dry period, as the detached weed will dry more rapidly. In moistweather the weed will often strike roots and keep on growing. Most toolsoperate better in sandy soil than in heavy clay soil, where the effect can bevery poor. Furthermore moist clay soil is restricted to drier periods. Rootweeds are difficult to control mechanically, whereas seed weed is easy tocontrol. The majority of the tools known today have not sufficient effect onstrong and well-developed weeds, bu t are effective against smaller weeds. For

many tools the effect is better speeds of 8-12 km/h than at 4-6 km/h, however,it can be difficult to drive at high speed without damaging the trees if the treeshave not been planted correctly at equal distances. The depth of soil tillagehas great influence on the effect on root weeds and larger seed weeds; deepsoil tillage can damage tree roots.

T o obtain a satisfactory result with mechanical weeding it is thereforeimportant to weed frequently, while the weeds are still small. One should notclean but keep clean. Blind harrowing before the weeds starts to grow often gives extremely satisfactory results. However, early actions can behindered by moist soil, which is not suitable for traffic and tillage. T he weedsstart to grow in April and grow tremendously from May and onwards. Sincethe risk of tools damaging the newly busted buds is large, it is advisable not todrive in the plantations for 2-3 weeks in the period of bud breaking. It istherefore essential to weed before bud breaking. Depending on thecircumstances the treatments have to be repeated 4-8 times per growingseason.

Only very few tools have the capacity to clean in the actual rows (spring tineharrow, hydraulic Dutch hoe, rotor harrow on flexible arm), most tools areonly able to weed in the inter-row area. In that case weeds are left around thetrees, where competition for water for the first 2-3 years probably is mostintensive. When the tree crown has grown bigger and denser the competition

for water will be most intensive in the periphery of the crown. The trees areby now shading away the majority of weeds near the stem, nevertheless there


16/86

16

will still be competition for light, and wear on the branches by tall weedspresent in the row.

3.2.2.1 Conditions for mechanical weedingThe planning is one of the most important conditions for a successfulmechanical weeding. The whole sequence of the cultivation has to bethoroughly considered and carefully planned. Firstly to make sure that theweeding practically can be carried ou t with the tools and machines at disposal,and secondly to ensure as low weeding costs as possible

Row length. It is crucial for the achievement and with this the economy ofmechan ical weeding, that the rows are as long as possible to minimise thenumber of turnings.

Appropr iate headland. It is importan t that the rows do not continue all theway out to any fence around the plantation. A headland sufficiently broad forthe later operat ing machines and tools is necessary. If the headland is toonarrow the operator of the machine is forced to go forward and backwards to

bring the machine in position.Irregular areas. Irregular areas should as far as possible be avoided or beadjusted, as wedge-shaped areas with the direction of row taperingproportional to the headland are time-consuming and space demanding toweed. It is comparatively time-consuming to turn and bring tractor and tool inposition. T he headland, which is an unproductive but indispensable area,must also be considerably wider for the tractor and tool to make turns.

Distance between the rows. T he distance between the rows must match thetools and machines to be used. It almost seems commonplace to mention this,but it is very often seen that the plan of cultivation had to be rearranged

because the distance between the rows appeared not to fit to the machines atdisposal.

Straight parallel rows. T o carry out mechanical weeding at a sufficiently highspeed the rows should be straight and parallel, at least in sets corresponding tothe width of the weeding implement. Sufficiently accurate planting can beobtained either by a planting machine with a number of rows which fit thesubsequent tools, or by drilling the area with a suitable number of holes beforethe plants are manually planted. Manual planting after two or more sticksdoes not give a satisfactory result.

The size of the weed. In general only a small number of the tools on themarket right now can manage taller weed. The tools normally operatesatisfactory on weed sizes up to 5 cm. If the weeds get much higher they maycause implement clogging or get entangled.

Soil conditions. Mechanical weeding is most successful and easy on light soil.The heavier the soil the harder mechanical weeding is to perform. On verystiff clay soils the traffic conditions in wet or m oist periods can be so d ifficult,as to exclude the p ossibility of mechanical weeding.

T he weather. T he ideal weather for mechanical weeding is warm, d ry and alittle windy. T he detached weed dries quickly and does not str ike roots again.

In p ractise it is not always possible to wait un til the weather is ideal because ofthe risk tat the weeds will grow too high. T herefore weeding is often carried


17/86


18/86

18

spade rollers are placed on two axles working towards each other. T he effectof the harrow depends on the travelling speed. It takes quite some speed 12km/h or m ore to obtain a satisfactory effect.

Hoe. A hoe is a tool that follows the rows. T he working organs can havedifferent shape, but the most common is tines with one- or two-wingedshares. If mounting half shares towards the row the roots of the trees will bespared damages. T he tool undercuts and covers the weed; the most importanteffect is the undercutting. T he tool controls weed efficiently; the hoe normallyeven has an effect on root weeds e.g. couch grass . Due to the aggressiveundercutting it has a good effect on higher weed, and is much less dependenton the size of the weed than the spring tine harrow. At any rate this alsoconcerns the hoe the less weed the easier the control. T he hoe can be usedthroughout the rotation. Only the height of trees compared to the tool bearersfree heightreduces its utility. The decisive disadvantage of hoe, is that it canonly weed in the spacing between the rows and not in the actual rows.

T he conditions of success with this type of tool:

that the r igidity of the tines fits the type of soil, that the winged shares give full cutt ing of the soil and have a good

overlap, that the tines,to give the highest possible material flow, are placed

with large distance that the tinesdo not go deeper than app . 5 cm, often even higher, that the travelling speed is at least 5 to 6 km/h, so that the tines are

vibrating, keeping themselves cleaned, and bringing the cut off ordetached weed to the ground surface.

Hydraulic Dutch hoe. T he hydraulic D utch hoe can be used for weeding inthe actual rows. A hydraulic run knife of 50-65 cm undercuts the weed at thedepth of 2-5 cm. The tool is mounted with a mechanically or ultrasonic sensorthat d isconnect the hydraulic system when it senses some resistance e.g. thestem of a tree. As the tool is driven forward due to the resistance from thesoil, the knife is pushed back and turns out into the space between the rows.After the tree has been passed the knife is then again hydraulically activatedinto the row. H ydraulic Dutch hoes are used in ecological as well asconventional fruit plantations. In the fruit plantations 7-8 haulages per seasonare needed to achieve sufficient effect. T he travelling speed is 7-10 km/h.

Rotary cultivator. The rotary cultivator is also an option. The knives aremounted on a horizontal-rotating axle. The rotary cultivator works by tearing

up, cu tting up and covering. T he effect on weeds is good, and the rotarycultivator can hand le even large amounts of weeds and thick grass. Rootweeds, however, can only be controlled with repeated treatments. The rotarycultivator can be problematic to use for repeated haulages in the cultivationdue to the risk of damaging the soil structure. The rotary cultivator is onlyweeding in the space between the rows. T he PT O run rotary cultivator has alow performance, whereas the friction driven rotary cultivator has a highperformance.

Under the category of rotary cultivators also belongs the hydraulic drivencultivator from the firm PolyT rac Inc. the so-called Mulcher. T heMulcher can tear up and break even high weed without getting entangled.

Rotary harrow. The rotary harrow has vertical stiff tine rotors that has a


19/86

19

cycloidal movement through the soil. T his levels the soil fairly well across thedirection of travel. T he firm of Silvatec produces a PT O driven rotary harrowintended for weed control in Christmas tree p lantations. T he rotary harrow isalso available mounted on a flexible arm, which makes it possible to weed inthe actual rows between trees big enough to reject the tool. The rotary harrowis suitable for weeding in overgrown cultivations.

3.2.3 AnimalsTogether with mechanical weeding grazing is among one of the mostpromising methods of weed control in Christmas tree cultivations at themoment. In principle grazing can be carried out with many different speciesof grazing animals; however, sheep of improved fattening breeds especiallyShropshire are the most used.

3.3 Har vest in g

The harvest of Christmas Trees starts in the early autumn with classificationand marking every tree. In the middle of November the actual harvestingstarts. Felling is accomplished using handsaws, brush cu tters or chain saws.During the past years there has been a development within mechanic fellingmachines.After felling the trees are dragged or carr ied out manually to thetractor tracks in the cultivation, where the trees are wrapped to protect themand facilitate further transportation to a central loading place easier. The treesare then loaded on lorries. Over the past years there has been a developmentof more mechanized transport systems, where the trees after wrappingareplaced on pallets, which makes the further transport considerably easier.

3.4 Cont r act or s and cont r act or cost s

T he use of contractors has become more common within Christmas treecultivations. This is partly because of the general reduction of the permanentstaff within the trade, and partly because the individual Christmas treeproducer gets the work done by experts having special machines at theirdisposal; machines that would not be economical for the individual Christmastree producers to own. T he contractor can today carry out all the tasks withinChristmas tree cultivation, from little isolated jobs to turnkey contracts.

It is difficult to get an exact overview of the contractor costs, as these arestrongly dependent on the conditions. Some contractors operate with a fixedrate per hour and a special road rate. Others have fixed prices per hectare withextra charges for small or difficult jobs/areas. Others again are not paid p er

hour or hectare but work out an offer for each contract

On good areas without road transport the contractor costs for mechanicalweeding with a tractor moun ted harrow are estimated to be between 2.200 to3.000 D KK/ hectare/year at a price per hour of about 400 DKK weeding withrotary cultivator/mulcher mounted at a special tool carriers will amount from1.250 to 2.850 DK K/hectare/year at a price per hour of about 475 DKK .depending on the p lanting and weeding system.

3.5 Costs of t he var ious oper at ions

In table 3.1 is shown a typical cultivation model for 1 hectare of Nordmanns

fir Christmas tree cultivation on former farmland. Planting is assumed to be


20/86

20

carried out manually, and spraying with leaf and soil herbicides is used forweed control.

Tabl e 3.1. Est imat ed cul t ivat io n cost s (DKK/hect are).Year from establishing 0 1 2 3 4 5 6 7 8 9 10 11 Total

Costs, DKK

Preparing chemical weeding 907 907

Soil preparation 1.155 1.155

Plants 13.889 1.389 15.278Planting including transportation 5.884 833 6.717

Fence 8.716 2.309 11.025

Chemical weeding 1.245 1.991 3.581 1.991 3.581 1.991 3.581 1.991 2.336 779 779 23.846

Insect control 679 679 679 679 679 679 679 4.753

Fertilisation 771 826 892 892 892 892 892 6.057

Pruning 200 500 1.000 1.000 1.000 1.000 1.000 5.700

Various 600 600 600 600 600 600 600 1.200 1.200 1.200 1.200 1.200 10.200

Cultivation costs total 32.396 4.813 4.381 2.591 5.452 4.096 6.752 5.762 6.107 4.550 4.550 4.188 85.638


21/86

21

4Weed control requirements inChristmas tree Production


T his chapter present results from an ongoing experiment that has the purposeof investigating the influence of different spatial weeding intensity onChristmas tree growth and survival on former agricultural land.

4.1 Method

T he species in the experiment are Nordmann fir (Abies nordmanniana). Twoparallel test areas were established; one in Randbl N ational Forest D istrict at T nballegrd, and one in W edellsborg Forest D istrict. The experimentalarea at T nballegrd is on a clay loam soil and the area at W edellsborg is on alight sandy soil. Planting was done with machines. Around each tree a circle isweeded m echanically.

The following treatments are included in the test (the percentage area iscalculated on the basis of a planting distance, which leaves a circular growthspace of app. 1.2 m 2 for each tree. The actual planting distance in the trial iswider):

1. 0% (un-weeded)2. 20% (mechanically weeded, r=28 cm or 0.25m 2)3. 40% (mechanically weeded, r=39 cm or 0.48 m 2)4. 60% (mechanically weeded, r=48 cm or 0.72m 2)5. 80% (mechanically weeded, r=56 cm or 0.99 m 2)6. 100% (mechanically weeded, r=62 cm or 1.21 m 2)7. 20% (chemically weeded)8. 40% (mechanically weeded planting p lace without p lants (only 10 circles

for T DR measurements))

T he trial is on singletree basis with 30 trees per treatment, for a total of 210

trees plus 10 empty planting spots. The trees are inspected for shootdevelopment at the start of each growing season. P lant survival, health, height,leader shot length, number of leader shots, number of branches in the topbranch whorl, root collar diameter are measured/registered after each growingseason.

In 10 circles of each treatment, the soil moisture content in the ploughinglayer (25 cm) is measured using the T DR method. In each circle the soilmoisture content is measured five times in increasing distance from thecentre: 15, 30, 45, 60 and 75 cm. T he soil water content is expressed in termsof percent of field capacity. Field capacity is determined by measurements ofthe soil water content in winter abou t 2 days after the last rainfall.


22/86

22

4.2 Resul t s

4.2.1 The average height of the trees.The average height of the trees can be seen in table 4.1 and 4.2.

Tabl e 4.1. Mean h eigh t of t r ees (cm), Tnbal l egaar d.

Treatment 1996 1997 1998 1999 20000 % 14,6 20,2 28,9 50,3 78,420 % 16,2 22,7 * 34,1 * 55,6 88,240 % 15,7 22,7 * 32,6 52,4 81,3

60 % 15,9 24,1 *** 35,9 *** 57,4 90,4 *80 % 14,6 22,0 33,4 * 51,5 81,0

100 % 14,3 22,8 * 36,3 *** 54,5 86,0

20 %chemicallyweeded

14,9 22,2 29,9 50,0 80,1

On the solid soil of Tnballegrd the trial shows that there has been nounambiguous effect of the treatment on the average height of the trees. T he

weeded trees have obtained a slightly higher average height, which in year2000 only in one single case is significant.

Tabl e 4.2. Mean h eight (cm), Wedel l sbor g.

Treatment 1996 1997 1998 1999 2000

0 % 11,1 13,4 17,8 28,1 37,320 % 10,8 16,7 * ** 21,6 ** 35,0 * * 49,3 **40 % 11,6 16,5 ** 22,9 ** * 37,0 *** 55,0 ***

60 % 11,2 16,5 ** 23,0 *** 36,3 *** 56,1 ** *80 % 11,1 16,2 ** 22,7 *** 36,2 * ** 53,5 ** *

100 % 11,6 16,3 * * 22,6 * ** 35,4 * * 54,8 ***

20 %chemically

weeded

10,3 14,5 18,3 31,5 42,8

In the Wedellsborg trial on the poor sandy soil the picture is considerablydifferent. From growing seasons 1997 to 2000 all the mechanical treatmentshad considerably larger height than the untreated (0 %). Apparently the 20 %treatment differs from the other treatments with a somewhat lower averageheight in year 2000, however, the d ifference is not significant.

The chemical treatment does not at any time differ significant from theuntreated. On the other hand the average height in year 2000 is significantlower then the 40, 60, 80 and 100 % treatments (p-values 0,0004 0,0001

0,0011 and 0,0001).


23/86

23

4.2.2 Root collar diameter of the treesThe root collar diameter of the trees can be seen in table 4.3 and 4.4.

Tabl e 4.3. Mean r oo t col l ar d iamet er (mm), Tnbal l egaar d.

Treatment 1996 1997 1998 1999 2000

0 % 6,7 10,9 16,2 23,2 29,7

20 % 7,3 13,0 * * 20,8 * ** 29,2 *** 35,9 **40 % 7,3 13,0 * * 20,5 *** 27,8 ** 34,1 *60 % 8,0 ** 14,5 *** 23,0 *** 32,0 * ** 38,8 ***80 % 7,8 ** 14,1 *** 22,4 *** 29,7 *** 36,2 **100 % 8,2 *** 14,2 *** 23,0 *** 31,0 *** 38,2 * **


7,6 * 12,2 * 18,4 * 25,2 33,2

It is evident that all the mechan ical treatments have given a significantly largerroot collar diameter from 1997 to 2000. T he 60, 80 and 100 % treatmentshave already even given a significant larger root collar diameter after the firstgrowing season. T he chemical weeded trees do not differ significantly fromthe untreated in the years 1999 and 2000.

Tabl e 4.4. Mean r oo t col l ar di amet er (mm), Wedel l sbor g.

Treatment 1996 1997 1998 1999 2000

0 % 6,4 6,1 9,2 11,0 14,8

20 % 6,1 8,0 *** 12,7 * ** 18,2 *** 23,7 * **40 % 6,2 8,6 *** 14,4 * ** 20,4 *** 26,6 ***

60 % 6,3 8,6 *** 14,7 *** 21,1 *** 27,7 ***80 % 6,1 8,3 *** 14,4 ** * 20,2 *** 26,5 ***100 % 6,0 8,4 * ** 14,9 *** 20,2 ** * 27,1 ***


6,0 7,6 *** 11,4 ** 15,0 *** 20,0 ** *

From 1997 and further on all the treatments have given significant larger rootcollar diameters than the untreated. In the year 2000 the chemical weededtrees have a significant smaller root collar diameter than all the mechanicaltreated (p-values 0,0151 0,0001 0,0001 0,0001 0,0001) .

4.2.3 Health score of the living trees.The state of health of the trees can be seen in table 4.5 and 4.6.

Tabl e 4.5. Tnbal l egaar d. Heal t h scor e of t he l iving t r ees (0 t o 10, wher e 0 = dead)

Treatment 1996 1997 1998 1999 2000

0 % 7,8 9,2 9,3 8,5 8,420 % 8,3 8,6 9,0 8,4 9,0

40 % 8,5 * 8,7 8,6 * 7,9 8,360 % 8,8 * * 9,1 8,5 * 8,1 8,780 % 8,3 8,9 8,3 ** 7,4 * 8,6

100 % 8,7 ** 8,5 * 8,6 * 7,9 8,9


7,6 8,3 * 8,9 8,0 8,4

It is evident that no immediate evidence of effects of the treatment on the stateof health of the trees on the rich soil. T here are significant manifestations in

the years 1996, 1997, 1998 and 1999. In 1996 an improved healthiness hasbeen obtained by weeding. The other years the health of the weeded trees hasbeen poorer than the health of the untreated trees.


24/86

24

Tabl e 4.6. Heal t h scor e of t he l iving t r ees (0 t o 10, where 0 = dead), Wedel l sbor g,

Treatment 1996 1997 1998 1999 2000

0 % 7,6 7,5 9,0 7,2 8,820 % 7,9 8,5 * 9,0 8,5 ** * 8,9

40 % 7,8 8,8 ** 8,9 9,0 *** 9,4 *60 % 7,7 8,7 * * 8,6 8,8 ** * 9,2

80 % 8,2 8,6 ** 8,8 8,8 ** * 9,4 *100 % 7,9 8,5 * 8,4 8,5 *** 9,3 *


7,4 8,1 8,6 8,3 * * 8,8

On the light sand soil there are significant effects of the treatments in the yearsof 1997, 1999 and 2000. In all cases there are positive effects due to theweeding.

4.2.4 Soil moistureSoil moisture is presented in the following figures as percent of the fieldcapacity. T his means that the curves will have a very characteristic course.Before the start of the growth season the moisture will be close to 100%,which is equal to the field cap acity. During the growth season the level will fallrapidly in the vegetation-covered areas and somewhat slower in the weededareas because of the evapo transpiration. In drought periods the level willequalise almost to the point of wither limit.

4.2.4.1 T nballegrdFigures 4.1 and 4.2 show the soil moisture in a normal and a dryer situation.

0

10

20

30

40

50

60

70

80

90

100

15 cm 30 cm 45 cm 60 cm 75 cm

Distance from plant

%o

ffieldcapacity

0%

20%

40%

60%

80%

100%

20 % Chemical

20 % No plant

Figur e 4.1. Soi l moi stu r e as % of f iel d capacit y. Tnbal l egrd 16/6 1998.

In the normal situation in figure 4.1 it can be seen, that there is a distincteffect of the treatm ents. T here is in the mechanically weeded trees a cleargradient, which reflects the effect of treatment, showing that the soil moisturehas a tendency to fall away from the weeded area close to the plant.


25/86

25

0

10

20

30

40

50

60

70

80

90

100

15 cm 30 cm 45 cm 60 cm 75 cm

Distance from plant

%o

ffieldcapacity

0%

20%

40%

60%

80%

100%

20 % Chemical

20 % No plant

Figur e 4.2 Il l ust r at ion of a ver y dry sit uat ion , where th e l evel pr obabl y is ver y cl ose

t o t he wit her l imit . There ar e no t seen any sign if icant dif f er ences in t r eat ment .

4.2.4.2 W edellsborgThe corresponding curves for the Wedellsborg trial can be seen in figures 4.3and 4.4.

0

10

20

30

40

50

60

70

80

90

100

15 cm 30 cm 45 cm 60 cm 75 cm

Distance from plant

%o

ffieldcapacity

0%

20%

40%

60%

80%

100%

20 % Chemical

20 % No plant

Figur e 4.3. Soi l moi stur e as % of f iel d capacit y. Weedel sbor g 21/6 1998.


26/86

26

0

10

20

30

40

50

60

70

80

90

100

15 cm 30 cm 45 cm 60 cm 75 cm

Distance from plant

%o

ffieldcapacity

0%

20%

40%

60%

80%

100%

20 % Chemical

20 % No plant

Figur e 4.4. Soi l moi stur e as % of f iel d capacit y. Weedel sbor g 21/8 1997.

T he gradient in the normal situation in figure 4.3 is much m ore distinct andreflects the effect of the treatment.

In the dry situation the level of moist in the mechanical weeded areas near theplants has come down to almost the same level as in the not weeded areas.

4.3 Ot her invest igat ion s

In each growing season the time of bud breaking, frequencies of leader shotsand mean number of branches in the top whorl were recorded. T heseparam eters, however, were not affected significantly by the d ifferenttreatments.

4.4 Discussion

Christmas Tree producers are basing their cultivation systems on traditionand practical experience, which has indicated that the Nordmanns fir isresponding to the weeding with better prosperity, growth and quality.Traditionally a 100 % weeding is therefore carried out in the Christmas treeplantations. In the trials reported here the trees have appeared to besurprisingly little influenced by the degrees of weeding above 40 %. T hus itcan be expected that fully satisfactory cultivation can be maintained with a

considerably reduced level of weeding.On both study sites the development of height has only been slightlyinfluenced by degrees of weeding above 40%. T he root collar diameter, whichnotoriously is a parameter very sensitive to treatment, on ly responds very littleon degrees of weeding above 40%. The general state of health of the trees hashardly been affected by degrees of weeding above 20%. Regarding thearchitecture of the trees the trials show that the number of trees with leadershot defects does not seem to be influenced by the degree of weeding, just asthe type of defect also seems to not be influenced . T his last issue is veryimportant as trees with more than one leader shot can be corrected by cuttingoff the extra leader shoots, and thus get into perfect shape. In contrast trees

with no leader shot at all are very difficult to repair. Also the number ofbranches in the branch whorl does not seem to be influenced by the degree of


27/86

27

weeding. The missing influence of the degree of weeding on the architectureof the trees is in good accordance with the fact that the time of bud breaking isalso not influenced by the degree of weeding in the tr ials. In this connection itis important to remember that the risk of spring night frost damages partlydepends on the time of bud breaking and partly on the locality. The trials areplaced on coastal areas and are thus not very likely to be exposed to springnight frost damages.

Regarding the soil moisture the T DR measurements from the trial shows thatunder normal circumstances in he areas with graduated mechanical weeding,there is more plant available water close to the trees than in areas withuntreated soil. T he T DR data also show that the mechanical weeding can beregarded as an insurance because soil drying out in drought situations occursmore slowly in the m echanical weeded soil than in the vegetation covered soil.

4.5 Sub con cl usio n

T o date, the he trials show that fully satisfactory cultivation results can beachieved on light as well as relatively heavy soils with area graduatedmechan ical weeding with a weeding degree of 40%, corresponding to aweeded circle shaped area of app. 0,5 m 2 around each plant.


28/86

28

5Specification of stakeholderrequirements

Frans T heilbyThe Danish Forest and Landscape Institute, Department of Forestry

Spyros Fountas and H enning NielsenThe Royal Veterinary and Agricultural University, Section forAgroTechnology

An important part of developing new technology is to define the desires orrequirements of stakeholders, i.e. the future growers, operators, manufacturers

and maintainers of the system. T hose users know the circumstances underwhich the system should work as well as the function and shortcomings ofpresent systems. They are also the people that have to live with the newsystem an d its properties in relation to people, work tasks and theenvironment.

To get a good basis for the development work a workshop was organised withthe project team and representatives from important stakeholder categories onthe 5th of March 2001. T he participants contributed with presentations onvarious aspects of Christmas tree production and on ideas in relation todevelopment of an autonomous Christmas tree plantation (ACW). After thisa brainstorm session was arranged to identify the current level of satisfactionwith spraying and m echanical weeding as well as a specification of stakeholderrequirements and wishes for an ACW. T he main outcome of this process isreported in the following.

5.1 The cur r ent l evel of sat isfact ion wit h spr ayin g and mechanicalweedin g t echn ol ogy

Weeding in the plantations is primarily done to reduce the direct competitionfor water, nutrients and light; however heavy weeds in Christmas treeproductions can also lead to wear damages, which are deteriorating to thequality of the trees. Furthermore, various weed species which are deep ly

rooted in the plantation can restrain or even totally ruin a Christmas treeplantation. In general the need for weeding is greatest in the phase ofestablishing the p lantation to secure survival.

Approximately 70% of the producers are using chemical weed control to someextent. Especially when using system 1, which mainly is used within fieldplantations, the u se of chemical control is widespread .

Among growers the general attitude is that it is desirable to introduce moreenvironmentally compatible weed con trol methods, if these also technically,economically and effectively are alternatives. T he use of chemical weedcontrol, especially the u se of chemical soil weed control, has been strongly

reduced in the recent years. T he improvement in spreading techniques, e.g.


29/86

29

band spraying and screened spraying, has also contributed to a reduction ofthe use of pesticides.

A political wish to reduce the use of pesticides has also contributed to theintroduction of various alternative weeding methods such as mechanical weedcontrol and weed control with animals. These methods are currently beingimproved and refined, but in general they are still too expensive. The cost ofthese alternative methods are 10-100% higher than spraying. At a time withdecreasing income per tree produced it is only natural if the producers aretrying to avoid increasing production costs, including weed control costs.

In general it is not the quality of the alternative weed control methods that isthe main problem, but the higher requirement of manpower that thesemethods imply.

5.2 St akehol der r equi r ement s to an ACW

In the p roject stakeholder requirements were specified on the basis of aworkshop with part icipants from the various stakeholder categories:

Requirements: Costs should be at most at the same level as chemical weed control Able to work on steep gradients and irregular soil surfaces Able to carry different tools, e.g. weed cutter and t illage tools Safe in relation to humans and animals Less than 5% trees damaged Simple ACW transport Easy maintenance ( standard spare parts) Secured against theft

Able to record important properties of trees and plantationWishes of additional work tasks:

Basal pruning Fertilizer application Spot spraying Tree marking for sale Shape regulation Growth regulation Transport of equipment for production measurements Recording of tree properties for management and on line sale

purposes Selective planting of new trees Selective tree felling (and possibly transport trees to a field handling

point) Selective rep lanting Spatially selective processing accord ing to needs.

In relation to the development stakeholders indicated that it is essential to getclarification about:

T he frequency of treatment T he appropriate periods of treatment (from the middle of May, the

beginning of July should be avoided for bud breaking reasons) T he possibility of combining different working operations.


30/86

30

5.3 Sub con cl usio n

The stakeholders found it desirable to introduce autonomous technology formechanical weeding in order to reduces environmental effects and labourrequirements, but this new technology should be competitive to the presentmethods. Also the stakeholders were interested in autonomous machines for anumber of other work tasks.


31/86

31

6Specification of technicalrequirements for an autonomousChristmas tree weeder

Henrik HaveThe Royal Veterinary and Agricultural University, Section forAgroTechnology


On basis of the described cultivation methods, minimum weed controlrequirements and stakeholder requirements the following full and reducedoperation scenarios and requirement specifications have been specified asgoals for the first generation autonomous Christmas tree weeder (ACW).

6.1 Over view of cul t ivat ion met ho dol ogy

Christmas trees are mostly grown on farmland in rows 90 to 125 cm apartand with similar intra row spacing. Another pattern is a 60 triangular pattern(referred to as rhombic), which corresponds to closest packing of equal sizecircles. Using this latter pattern about 20% more trees can be grown on thesame area. Different tree species are chosen depending on the soil types.

The production cycle starts when typically two year old trees (about 20 cmhigh) are p lanted on prepared land, and it ends after 6 to 10 years when thetrees are cut selectively. In some cases new, small trees are planted in betweentrees remaining from this selective harvest.

Mechanical weeding during the first two years after planting is normallyperformed with special spring tine harrow (langfingerharve) developed forweeding in agriculture (see chapter 4), which is relatively cheap, and yields asatisfactory result. At the later growth stages this method is not suitable. T hisis when the ACW is to be employed.

Weeding is needed until the trees are about 1,5 m high. It must at least cover acircular area of 40 cm radius from the stem to avoid competition and stiffshoots of perennial weeds (including trees) to cause mechanical damage to thebranches.

T he weed control may be performed either as shallow soil tillage or as weedcutting near the ground. T he latter provide sufficient reduction of weedcompetition, but in some areas it is an advantage to clear the soil around thetrees during springtime, as it reduces the r isk of frost damage of emergingshoots. A shallow form of soil tillage is considered suitable for this purpose.Weeding should be done 4 8 times a year (chapter 4). Recently emerged

shoots are brittle and easily damaged mechanically.


32/86

32

About 6 years from planting, when the trees are about 1 m high, some of thelower branches of the trees are removed to reduce the top shoot growth. T hisoperation is in some cases done twice. Further growth regulation is sometimes made by removing part of the bark around the lower part of the stem.

Late in the growing period tracks may be cleared in the plantations fortransport depending on the harvest strategy (chapter 3) .

6.2 Def in it io n o f oper at ion scenar io s f or t he ACW devel opment

It is clear that the full range of plantation scenarios occurring in practicalconditions is wide, and that it would be difficult and costly to develop anautonomous machine that could cope with all situations. Therefore asomewhat narrower range of scenarios is chosen for the analyses.

Tabl e 6.1. Chr ist mas pl ant ati on paramet er s and r equ ir ement s chosen f or t he ACWdevel opment .Category Parameter Occurring conditions Chosen machine requirements

Unevenness < 5 cm from average < 5 cm from average

Amount and type ofresidue First generation: Noresidue.Second generation:Branches and stubbles

None: Residue to be clearedbefore second generationplanting

Inclination < 12 %

at the branch tip ->

Depends on speciesand age

020 cm

030 cm

Min. 5 cm

Tree size,and

shape

Trunk diameter 3 8 cm 3 8 cmTemperature -5 to + 35 -5 to 35ClimatePrecipitation All Resistance to rainAnimals Dears

Nests and youngstersAvoid stationary items greaterthan 10 cm

Uncon-trolledobjects People Curiosity, larceny Stop, send alarm

In addition the mechanical damage of trees between planting and harvest

should be less than 5%. Also any contact between trees and the ACW shouldbe gentle or avoided during 3 weeks from medium May when the bud arevery sensitive to damage.

6.3 Def init ion of wor k t asks

During the workshop and further analyses (chapter 5) the followingoccurring as well as new work tasks were identified as suitable for one ormore autonomous machines. However it was decided to limit this initial workto weeding while keeping the other potential tasks in mind (Table 6.2).


33/86

33

Tabl e 6.2. Wor k t asks consider ed in t he pr esent anal ysis.Work tasks Main parameters Possibilities ChoiceMethod Weed cutting, tillage and

othersWeed cutting

Tool Rotary cutters, drum cutters Rotary cutter withhinged exchangeableknifes

Weed development stage Early to late Relatively early

Weedcontrol

Frequency of operation Four to eight Ad-hocFrost riskreduction

Removal of weeds andtrash around trees

Usual cult ivation method.ACW with tillage tool.

Usual cultivationmethod

6.4 Sub concl usion

In conclusion it is recommended to focus on development of an autonomousvehicle for weeding with rotor cutter near the ground in a slightly reducedrange of occurring scenarios. Further development of the vehicle for some ofthe other tasks appears feasible, but should be postponed to a later stage.


34/86

34

7Autonomous vehicle technology a literature review

Henning Nielsen and Spyros FountasThe Royal Veterinary and Agricultural University, Section forAgroTechnology

Autonomous vehicles (also called mobile robots and robotic vehicles) arenormally categorized as developments from automatic steered vehicles andremote controlled vehicles, while the term robot usually is used for materialshandling and tool operating automata mostly used in industry. Varieties of thetwo have found use in agriculture. An example is the experimental, robotic

fruit harvesters developed years ago, which moved (autonomously) from treeto tree and picked the fruit by mean of an robotic arm. Another example is themilking robots, which are used in p ractical farming, milking the cows whenthey prefer. Human like robots ought to be regarded as an entertainmentindustry phenomena.

The most primitive robots, sometimes called industrial robots are just materialhandlers, mostly pneumatic, typically running through a process of gripping awork piece, carrying it and placing it in another place and returning to startthe next cycle. True industrial robots are versatile equipments able to be setup to perform work tasks like seam welding, painting, moving a work pieceand assemble components. T o do this it should be reprogram able in an easyway without physical changes, have a memory and logic to be able to workindependently and automatically and further have a physical structure of afashion that allows for its use for several tasks without m ajor restructuring(Lundquist, 1996).

Another area relevant to mention about work on robotics and au tonomousvehicles is toys. T he LEG O Mind stones products have by themselvescontributed to the development, but the products have also been found usefulas experimental modelling tools. Works like Rooker & Lund (2001) may beapplicable for developing the man-machine interface for the operators settingup the work of autonomous machinery. This reference describes a

programming tool for LEG O robots to be used by children for setting upautonomous robot soccer players.

Autonomous m ini sub-marines designed for under-sea prospecting and searchare mentioned by different sources, their relevance for the actual purpose is

judged to be limited.

Going into the real life of designing autonomous vehicles, Gomi (2001) statedthe priorities of hardware units to be: Battery, motor, connector, sensor, cpu.

The greatest future application of autonomous vehicles, many with roboticactions, is expected to be for domestic purposes including assistance todisabled people (Christensen, 2001; G omi, 2001). At present most work inthis area is at the p re-commercial stage. An exception is a vacuum cleaner


35/86

35

from Electrolux. Although the research on other of these types of equipmenthas reached a rather high level of perfection, marketing for end users mustprobably wait for some time for safety reasons. However there is a great rangeof commercial vehicle types available as platforms for research purposes(figure 7.1). Some of these contain controllers with great computing capacity.

Figure 7.1: Example of skid steered commercial available autonomous vehicle platform for

experiments. The unit is electric powered from batteries and fitted with basic sensors, computersand communication equipment, which are open for addition of supplementary units.

7.1 Aut omat ic and aut on omous veh icl es in agr icul t ur e, f or est r y andhor t i cu l t u re

Petersen ( 1985) described some p ossibilities for application of robots inagriculture and mentioned p icking of fruit, harvesting of vegetables as well astransplanting and spraying. Research has been reported on robotic harvest ofapples, grapes and oranges (Burely et al. 1990), robotic harvest of apples(Kataoka et al. 1997), robotic pruning of grapes (Lee et al. 1994) and robotic

harvest of vegetables (Hilton, 1997).

Kondo & T ing (1998) describes a number of robots for fruit and vegetableharvesting. Robots for other crop growing operations are also shown. Many ofthe robots are mounted on vehicles, some without an operator. Apparently thedescribed robots rank from commercial to experimental. T he treatment showsthe importance of sensors as part of the robots. Reports related to roboticweeding, e.g. Molto (1997) m ostly consider weed sensing, in most cases usingvision systems. According to Kondo & Ting (1998) the interest for robotictractors (~ autonomous tractors) in Japan is increasing.

Except for some robotic equipment mentioned above the only autonomous

vehicle marketed for agriculture-related purposes seems to be lawn movers.Most well-known are two types sold by Husquarna, one solar poweredanother with battery loading from a mains connected servicing station that themachinery drive to when needed . T hese lawn movers move the grass in arandom linear pattern like the abovementioned vacuum cleaner formElectrolux.

Autonomous vehicle platforms for experiments are commercially available,e.g. from Applied AI Systems, Inc. (2001)(figure 7.1). This firm offers anumber of different types (called robots), mostly for indoors use; a few can beused outdoor. The problem is that the latter types are skid-steered, which isnot very suitable for weeding purposes. Up to now no commercial vehicleplatforms, which are easy to adapt for development of an autonomous weederhave been found.


36/86

36

An apparently more suitable vehicle platform has been made by Madsen &Jacobsen (2001) (Figure A.2, appendix A) as their M .Sc. thesis project. T hisis a robotic vehicle designed as a carrier for weeding purposes. The vehiclehas four-wheel drive four-wheel steer and is designed by mechatronicsprinciples with individual steering motors at each wheel

7.2 Evol uti on o f aut omat ic st eer ing f or agr icul t ur al vehicl e

Experiments on ideas of removing the tractor driver or easing his steering jobhave already been reported from 1909 (AG RI/WP.2/69, 1962). Since thenthe technology has developed to make automatic steered vehicles possible. Asstated in a review by Wilson (2000) the fact is that for automatic steering ofagricultural vehicles of today the guidance will in most cases use GPS (G lobalPosition System) for absolute position sensing and vision systems (imageprocessing systems) for relative position sensing. A further element necessaryfor practical automatic steering are controllers based on computers withreasonable computing capacity.

T he history of autom atic steering of agricultural vehicles can briefly bedivided in two epochs. Until about 1940 the experimens had mostly been onmechan ical systems. A p rominent result was furrow followers delivered asstandard equipment for some tractors. Remote radio control of tractors wasdemonstrated in 1936. D uring the second world war servomechanisms wereput into general use. This developed the theory of control systems.

From the 1950es to the 1980es a great amount of research on automaticsteered agricultural machinery was performed. Reviews of this is found in thereferences above and in Jahns (1976) and Nielsen et al. (1976). A great partof this research considers sensor principles for guidance information. Further

research considers design of the control systems, some on m ore generallyapplicable principles and mostly on detailed electronic design with minorrelevance today. Liljedahl et al. (1962) applied the theory of automatic controlon automatic tractor steering and by this introduced application ofmathematical modelling or systems analysis into the subject.

In the 1970es the increased application of electronic instrumentation foragricultural engineering research was crossbreed with electronic control forthe advance of both topics, and at the end of the decennium themicroprocessor was introduced in some systems. The decade also saw the firstsimple electronic monitoring instruments in practical farming.

During the 1980es electronic equipment found more widespread use inpractical farming, mostly monitoring equipment and some simple controls. Inthis decade also targeted work to increase the quality of farm electronics hadbeen performed. D evelopment and standardization of data bus systems foragricultural tractors were also started up in th is period.

In the 1990es the concept of precision farming brought positionmeasurement (~ navigation systems) into p ractical farming with G PS (G lobalPositional System) in practice becoming the universal sensor for absoluteposition. To obtain sufficient accuracy error correction using differential GPShad to be used, but since the intentionally introduced error (selective

availability) has been removed other m eans of obtaining accuracy havebecome possible. However, for guidance purposes the accuracy obtained by


37/86

37

these systems is not sufficient. In stead RT K GPS ( Real T ime KinematicGPS) has to be used. T he RT K system extends the differential principle withcorrections based on application of the GPS carrier wave phase.

T he 1990es also saw the development of image processing systems able toprocess relatively comp licated images nearer to real time. T hese systems werepartly developed to become research measurement tools, partly researched forpost harvest processing of horticultural products (Bennedsen et al., 1996;Bennedsen & Kohsel, 1996; Bennedsen, 1997). Some of the latter have beendeveloped into systems, which now are inserted into p roduction lines inhorticulture (Anonymous, 1997). Fur ther research has been made onapplication of image p rocessing systems for ap plication by control of fieldmachinery, e.g. fruit picking (Peterson & Bennedsen, 1999), weed detection(Pedersen, 2001) and guidance in row crops. T he latter has been developedinto control systems marketed for guidance of row crop cultivators(Bundgaard, 2001).

An updated broader overview on automatic steering of farm vehicles can be

found in a thematic issue (no. 25, 2000) of Computers and Electronics inFarming.

7.3 Sensor s f or navigati on

The primary sensors needed for navigation are systems that can provideinformation on absolute and relative position, vehicle absolute and relativeorientation and speed. Supplementary systems may be needed for specificpurposes. Many other sensor types may also be used internally in machineryand for crop sensing, e.g. tactile sensors for mechanically sensing presence ofmaterial or force from material. It is worth noting that some in generaloutdated guidance principles have properties, which may make them

candidates for special applications.

7.3.1 Real time Kinematic global position systemAs mentioned above navigation (~guidance) of automatically steered vehiclein future is today assumed mostly to be based on measurement of the absoluteposition based on RTK GPS (Real time kinematic global position system)and relative position determined by vision.

T he RT K G PS is based on measurements of the propagation time for theradio waves from the Navstar satellites. The RTK principle is based onapplication on a local ground based reference station and obtains centimetrelevel accuracy using the carrier wave phase difference. T he principles aredescribed in a number of textbooks and articles. A short technical descriptionof an actual system is, e.g. found in a manual from T rimble (1999). As aminimum a GPS equipment will deliver horizontal coordinates in a groundbased coordinate system, but most provide also the vertical coordinate, timeand additional information, e.g. about accuracy.

7.3.2 Computer visionT his technology is in many cases based on use of video cameras, oftencommon colour cameras (giving RGB colour information) or grey level

cameras. In some cases cameras with other spectral sensitivities or even morespecial camera types are used. T he camera signals are after digitisingprocessed computationally to extract relevant sensor information, e.g. the


38/86

38

vehicle heading relative to a crop row. A great am ount of literature existsabout vision systems and image processing. A part of this is about vision forguidance purposes.

7.3.3 Supplementary sensorsBesides both the above primary navigational sensor systems reliable guidance

often will depend on supplementary sensors, among others to detect theangular orientation/attitude of the vehicle. This comprises the threeparameters:Heading, roll andpitch. The heading is the direction of drivingexpressed as an angle relative to another direction; it is primary informationfor control of the steering. Absolute measure for heading is the compasscourse. D ifferent com pass sensors are available, but the heading can also becalculated from a number of subsequent position measures when driving. Rolland pitch are the angular deviation of the vehicles vertical axis from the actualvertical, roll is the sideward angle and pitch is the up-down angle of thevehicles forward direction. Both can be measured with inclination sensors.For accurate position determination with RT K G PS roll and pitch are often

needed for correction because the GPS aerial is placed at a higher level thanthe tool to be positioned.

In some cases information on driving distance and velocity are needed for agood control of vehicle driving. This can be supplied by an odometer. Formeasurem ent of shorter distances or as an alternative to vision, sensors basedon ultrasound can be used.

7.3.4 The leader-cable principleBy this principle the vehicle follows a single cable buried below the track. T hecable carries a voice frequency alternating current. The approach is

commonly used for automatic transport systems in industrial plants, where thedriver-less vehicles have bumpers operating safety stops to avoid damage onpersons or other objects in the vehicle path . Variants of this system have beenused or been tried for quite a lot of applications. A system for farm use wasdescribed by M organ (1958) and since then more advanced experimentalsystems have been described, e.g. Jahns (1976,b). Bearing some resemblanceto the present project, is the application for automatic lawn movers where thecable signals the limit of the area to be m oved. Some autom atic movers alsouse a leader cable for homing guidance to a servicing station. In a GPS-basedsystem a leader cable could be used as part of an emergency subsystem.

7.3.5 Laser-based principlesT he use of lasers has possibilities for high accuracy in areas where line of sightcan be obtained. U sing a laser beam as a very straight guidance line is used fordifferent purposes. Lasers can be used for accurate measurement of angles inone or two dimensions by mechanically scanning the beam in a plane angle ora solid angle respectively. Further the distance can be measured by means of amodulated laser beam reflected back into the instrument with the laser.Distance measurement and angle scanning is sometimes combined in oneinstrument.

A commercial version of this combined principle made for accurate guidancein limited areas is described by Arnex (1995) and Sgaard (1998). T hesystem has a laser scanner on the vehicle and uses reflective fix-pointsinstalled around the field in positions accurately surveyed. T he system delivers


39/86

39

two measurements per second for each reflector. Because of the lowmeasurement frequency the equipment has a Kalman filter, which is asoftware estimator to extrapolate into the time interval between lastmeasurement and the time when next position coordinates have beenmeasured and calculated. Another remarkable feature is optics to broaden thelaser beam so the light intensity is reduced to eye-safe level.

Laser scanners and laser distance-meters may be used as test instruments.Kalman filtering is generally applicable for real time purposes e.g. in controlsystems with t ime discrete and noisy signals. Laser scanners m ay also beapplicable for replacements of vision systems, especially if range data arerequested for 3d imaging.

7.3.6 Other principlesOther relative position sensors with p ossible relevance are mechanical typeswith a feeler arm. For instance the type used for sensing maize stalks(Kutzbach & Quick, 1999; Suggs et al., 1972), could be modified for sensing

Christmas tree trunks.Simple optical sensors sensing presence or non-presence of material havebeen used for many purposes, but not very many have been used in fieldmachinery because of the risk of contamination.

Simple optical sensors with ability of discriminating between green plants andother materials have been used experimentally in the 1970es (Palmer &Owen, 1971; Hooper et al., 1976). The special relevance is that these sensortypes, instead of using the green colour used the more pronounced differencebetween red reflectance and near-infrared reflectance of green plant materials;this feature is also used in some modern vision-based systems.

T illet (1991) has produced a more recent review of automatic guidancesensors for agricultural field machines.

T he topic degree of autonomy has been discussed by C astelfranchi (2001) .Degree of autonomy is approximately the same as level of delegation. Thetopic brings in a lot of implications partly of rather philosophical nature.

7.4 Safet y

Safety of machinery working without continuous human supervision iscritical. Apart from the functional reliability the machinery must not hurt

human or animals and should neither damage third persons property. Forunmanned machinery working in open country the p roblem of possible entryby unwanted persons into the work area is serious.

The controller must handle safety problems as part of its operation. In thecase of damage to the machinery or other error conditions the design of thecontroller and the rest of the machinery must assure a safe function with agraceful degradation.

For the matter of safety of third parties and the operator an alternative outerlayer of safety must be provided. In most cases this will be performed by ahardware safety system independent of the controller. In particular themachinery must stop if children or animals are coming near, but not if theoperator or another authorized person is present in a safe way.


40/86

40

The safety systems must comply with the rules set up by safety at workauthorities. A first approach for safety of an area with working machines isfencing with safety switches stopping the machines in case the fence isopened, a solution which may be prohibitively costly in the case of weedingmachinery and also a problem in relation to the work of an operatorsupervising the work now and then. In general if a specific machine is notcovered by special rules the safety system of an actual machine will be judgedto comply or not with a generalized set of rules. Danish rules possiblyapplicable to the safety of autonomous field machinery are at least partlycovered by the rules for remote control of technical equipment(Arbejdstilsynet, 1995a) and the rules for automatic controlled machinery,industrial robots included (Arbejdstilsynet, 1995b).

An elementary set of safety demands to automatic steered agricultural vehicleshave been set up by Jahns (1975).

7.5 Syst ems Ar chit ect ur e

An architecture is a description of how a system is constructed from basicsand how those components fit together to form the whole (Albus, 1991).

Mobile robots, if they are to perform useful tasks and become accepted inopen environments, m ust be autonomous: capable of acquiring informationand performing tasks without programmatic intervention. D ue to thecomplexity and intelligence of an au tonomous vehicle it is necessary toincorporate systems architecture already in the design phase. As a result, theliterature review is derived by the Artificial Intelligence and Robotics.

T here are not many references about systems architectures for autonomous

vehicles in agriculture. N ilsson (1980), states that a control system for anautonomous tractor should be decomposed into three functional elements: asensing system, a planning system, and en execution system. The job of thesensing system is to translate raw sensor input into a world model. The job ofthe planner is to take the world model and a goal and generate a plan toachieve the goal. T he job of the execution system is to take the plan andgenerate the actions it prescribes.

Later on, the sense-plan-act approach (SP A) became the dominant one in thisarea. The SPA approach has two significant architectural features. First, theflow of control among these components is unidirectional and linear. Second,the execution of an SPA plan is analogous to the execution of a computer

program. Executing a p lan or a program is easy when compared withgenerating one. T herefore, the intelligence of the system lies on the p lanner orthe programmer and not on the execution mechanism (Connell, 1989).

T he next step of the sense-plan-act (SPA) app roach was the subsumptionapproach, applying task-dependent constraints to the subsumption layers tomake SPA more efficient. The most well known example of this approach isthe mobile robot called Herbetts which was programmed to find and retrievesoda cans in an office environment (Connell, 1989).

Rzevski (1995) mentions three main types of three systems architectures for

mobile robots. T he hierarchies, networks and layered architectures.


41/86

41

A hierarchy is an architecture that consists of elements linked as p arentsand children . Each parent can have one or more children. Each child maybe a parent of other children. In this way multilevel hierarchies areconstructed. Hierarchies are used whenever it is necessary to reduce theperceived comp lexity of a system caused by its scale (size).

In networks, in contrast to hierarchies, there are no levels of importance andall elements may be connected to each other. These architectures are usedwhen there is a need for cooperation be

Autonomous Weeders for Christmas Tree Plantations

Documents

Transcript of Autonomous Weeders for Christmas Tree Plantations