developing afuzzy logic controlled agricultural...

7
AUTONOMOUS VEHICLES Developing a fuzzy logic controlled agricultural vehicle Hani Hagras, Victor Callaghan, Martin Colley and Malcolm Carr-West Malcolm Carr-West Abstract This paper describes the design ofafuz...,)· controlled autonomous rohot for use in an outdoor agricultural environment for cropfollowing. The robot has to navigate under different ground and weather conditions. This results in complex problems of identification, monitoring and control. III this paper. a [u controller is identified which when used in conjunction with a novel outdoor sensor design deals with both crop tracking and cutting. The controller was tested on an indoor mobile robot using two ultrasound sensors. The controller showed C/ good response in spite of the irregularity of the medium as well as the imprecision in the ultrasound sensors. The same controller was then transferred to both an electrical and diesel powered robots which operate in all out-doorfarm environment. These outdoor robots have used our novel sensor (mechanical wands) as well as outdoor ultra sound sensors. 177erobot had been tested in outdoor environments onfences and real Hani Hagras, Victor Callaghan and Martin Colley are at The Computer Science Department, Essex University, Wivenhoe Park, Colchester, England, UK; and Malcolm Carr-West is at The Agricultural Engineering Department, Wri"'e College, England, UK. 10 crop edges. The robot displayed a good response following irregular crop edges full of gaps under different weather and ground conditions with ill a tolerance of roughly 50 1Il/11. 1. Introduction The problem of a decreasing agricultural workforce is universal. Therefore, there is a need for automated farm machinery, ultimately including unmanned agricultural vehicles. Many machinery operations in agriculture are essentially repetitive and work with crops planted in rows or other geometric patterns. These operations involve making a vehicle drive in straight lines, turn at row ends and activate machinery at the start and finish of each run. Examples of this are seen in spraying, ploughing and foraging. In agriculture, the inconsistency of the terrain, the irregularity of the product and the open nature of the working environment result in complex problems of identification, control and dealing with sensing errors. These problems include dealing with the consequences of the robotic tractor being deeply embedded into a dynamic and partly non-deterministic physical world (e.g. wheelslip, imprecise sensing and other effects of varying weather and ground conditions on sensors and actuators). Fuzzy logic excels in dealing with such imprecise sensors and varying conditions which characterises these applications. Artificial intelligence (AI) techniques including expert systems and machine vision have been successfully applied in agriculture. Recently, artificial neural network and fuzzy theory have been utilised for intelligent automation offarm machinery and facilities along with improvement of various sensors. Ziteraya and Yamabaso (1987) showed the pattern recognition of farm products by linguistic description with fuzzy theory was possible. Zhang et al, (1990) developed a fuzzy control system that could control maize drying. Ollis and Stentz (1996) has used machine vision to follow and cut an edge of a hay crop but he did not address the problem of turning around at the end of bouts or the detection of the end of a crop row. Cho and Ki (1996) has used a simulation of a fuzzy unmanned combine harvester operation but he used only on- off toucb sensors for his fuzzy systems and hence lost the advantage of fuzzy systems in dealing with continuous data which had led him not to have smooth response and gave him problems when turning around corners. It must also be noted that all of his work was in simulation which is different from tbe real world farm environment. Yamasita (1990) tested the practical use of fuzzy control in an unmanned vehicle for use in greenhouses. Mandow et al. (1996) had developed the greenhouse robot Aurora, but the application and environment variation in greenhouses are more restricted and controlled than those in the field. Little work hall been done in implementing a real robot vehicle using fuzzy logic which can operate in the open field. The aim of this paper is to develop a fuzzy vehicle controller for real farm crop following. An emulation of 'crop- Landwards, Summer 2000

Upload: others

Post on 22-Sep-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Developing afuzzy logic controlled agricultural vehiclevictor.callaghan.info/publications/2000_JAE00(Developing...Fuzzy logic excels in dealing with such imprecise sensors and varying

AUTONOMOUS VEHICLES

Developing a fuzzy logiccontrolled agriculturalvehicleHani Hagras, Victor Callaghan, Martin Colley andMalcolm Carr-West

Malcolm Carr-West

AbstractThis paper describes the design ofafuz...,)·controlled autonomous rohot for use inan outdoor agricultural environment forcrop following. The robot has to navigateunder different ground and weatherconditions. This results in complexproblems of identification, monitoringand control. III this paper. a [u:»controller is identified which when usedin conjunction with a novel outdoorsensor design deals with both croptracking and cutting. The controller wastested on an indoor mobile robot usingtwo ultrasound sensors. The controllershowed C/ good response in spite of theirregularity of the medium as well as theimprecision in the ultrasound sensors.The same controller was then transferredto both an electrical and diesel poweredrobots which operate in all out-doorfarmenvironment. These outdoor robots haveused our novel sensor (mechanicalwands) as well as outdoor ultra soundsensors. 177e robot had been tested inoutdoor environments onfences and real

Hani Hagras, Victor Callaghan andMartin Colley are at The ComputerScience Department, Essex University,Wivenhoe Park, Colchester, England,UK; and Malcolm Carr-West is at TheAgricultural Engineering Department,Wri"'e College, England, UK.

10

crop edges. The robot displayed a goodresponse following irregular crop edgesfull of gaps under different weather andground conditions with ill a tolerance ofroughly 50 1Il/11.

1. IntroductionThe problem of a decreasingagricultural workforce is universal.Therefore, there is a need forautomated farm machinery,ultimately including unmannedagricultural vehicles. Manymachinery operations in agricultureare essentially repetitive and workwith crops planted in rows or othergeometric patterns. These operationsinvolve making a vehicle drive instraight lines, turn at row ends andactivate machinery at the start andfinish of each run. Examples of thisare seen in spraying, ploughing andforaging.

In agriculture, the inconsistency ofthe terrain, the irregularity of theproduct and the open nature of theworking environment result incomplex problems of identification,control and dealing with sensingerrors. These problems includedealing with the consequences of therobotic tractor being deeplyembedded into a dynamic and partlynon-deterministic physical world (e.g.wheelslip, imprecise sensing andother effects of varying weather andground conditions on sensors andactuators). Fuzzy logic excels indealing with such imprecise sensors

and varying conditions whichcharacterises these applications.

Artificial intelligence (AI) techniquesincluding expert systems and machinevision have been successfully applied inagriculture. Recently, artificial neuralnetwork and fuzzy theory have beenutilised for intelligent automation offarmmachinery and facilities along withimprovement of various sensors. Ziterayaand Yamabaso (1987) showed the patternrecognition of farm products by linguisticdescription with fuzzy theory waspossible. Zhang et al, (1990) developeda fuzzy control system that could controlmaize drying. Ollis and Stentz (1996) hasused machine vision to follow and cut anedge of a hay crop but he did not addressthe problem of turning around at the endof bouts or the detection of the end of acrop row. Cho and Ki (1996) has used asimulation of a fuzzy unmanned combineharvester operation but he used only on-off toucb sensors for his fuzzy systemsand hence lost the advantage of fuzzysystems in dealing with continuous datawhich had led him not to have smoothresponse and gave him problems whenturning around corners. It must also benoted that all of his work was insimulation which is different from tbe realworld farm environment. Yamasita(1990) tested the practical use of fuzzycontrol in an unmanned vehicle for usein greenhouses. Mandow et al. (1996)had developed the greenhouse robotAurora, but the application andenvironment variation in greenhouses aremore restricted and controlled than thosein the field. Little work hall been done inimplementing a real robot vehicle usingfuzzy logic which can operate in the openfield.

The aim of this paper is to develop afuzzy vehicle controller for real farm cropfollowing. An emulation of 'crop-

Landwards, Summer 2000

Page 2: Developing afuzzy logic controlled agricultural vehiclevictor.callaghan.info/publications/2000_JAE00(Developing...Fuzzy logic excels in dealing with such imprecise sensors and varying

AUTONOMOUS VEHICLES

Fig. J The indoor robot and itssensor configuration

following' (which is also an example offence following) is presented and itsresponse and control surfaces areanalysed. Then the same control

2. Problem definitionIn this section, we introduce thearchitecture of the robot and describe ournovel sensor design which is suitable forsensing crop boundaries. The robot isdesigned to mow a crop by following itsedge while maintaining a safe distancefrom the uncut edge, in this case 450 mm.While the development work has beenbased on mowing, the team have takeninto account the requirements of otherfieldwork operations.

Initially, we have tested our designwith an indoor mobile robot, introducingto it all the hard conditions that it mightencounter in a real field. Although thereare clearly big differences between theindoor environment and that in the field,

Fig. 2 (a) The outdoor electrical robot; and (b) the outdoor diesel robot

architecture was moved toour outdoor robots, Theserobots are equipped withspecial outdoor sensors (amechanical wand and anoutdoor ultra sound sensor)which are designed to dealwith the crop characteristics.The fuzzy controller hassucceeded in followingvarious outdoor crop andfence edges ranging frommetal structures. lines oftrees, to crops of hay(including irregular edgeswhich include small gaps)within a tolerance of 50 mm.It has shown it ability to turndifferent kinds of cornerssmoothly and worked in avariety of weather andground conditions.

Landwards, Summer 2000

r Knowledge 1

1 base

Fuzzy Fuzzy

[ Fuzzification 1---.1 Decision making~

Defuzzificationinterface logic interface

Controlled system(process)

Actual control Actual control(non fuzzy) (non fuzzy)

we have done what we could to make theexperiments more realistic, such as usingnoisy and imprecise sensors, irregulargeometrical shapes and fencesconstructed from baled hay. However, itis-self evident that ultimate test of a farmrobot is in the field and we thus includedas a subsequent stage an assessment basedon the use of our outdoor electric anddieseL-vehicles. We feel that thisapproach is better than a computersimulation which suffers from wellknown modelling difficulties (especiallywhen trying to model the physicalenvironment comprising varying groundand weather conditions and objects suchas trees telegraph poles).

2.7. The robot descriptionThe diesel field robot is constructed onthe chassis of a three wheel farm truck.The engine is a New Holland threecylinder diesel engine coupled to ahydrostatic transmission to thedifferential. Hydrostatic transmissionwas chosen as providing a simpleclutchless transmission. Steering poweris provided by an auxiliary pump to a nonbalanced double acting ram. This waschosen as providing a more complexcontrol problem tban a balanced ramwould have done. The electric outdoorrobot is about the size of a wheelchairand indeed utilises many wheelchairparts. Both robots have mechanicalwands (potentiometer arms connected toanalogue to digital converter to sense theedge of a crop), ultra-sound sensor, global

positioning system(GPS), and a camera.The camera forms partof a system developedby our group (Scballter,1996) to locate haybales. The electricrobot have two separatemotors for traction andsteering.

The indoor robotshown in Fig. 1has aring of seven ultrasonicproximity detectors, an8-axis vectored bumpswitch and an infrared(JR) scanner sensor toaid navigation. It alsohas two independentstepper motors fordriving the frontwheels, with steeringby driving at different

Fig. 3 The basic configuration of a fuzzy logic controller

11

Page 3: Developing afuzzy logic controlled agricultural vehiclevictor.callaghan.info/publications/2000_JAE00(Developing...Fuzzy logic excels in dealing with such imprecise sensors and varying

AUTONOMOUS VEHICLES

motors speeds. We try to giveall our robots a similararchitecture (to simplifydevelopment work) so itshardware is also based onembedded Motorolaprocessors (68040) runningVxWorks RTOS.

Other papers reportedproblems using certain typesof sensor in outdoorenvironments. One reportedsolution uses simple touchsensors (Cho &Ki, 1996)which have ON-OFF stateson ly which is not efficient forfuzzy control. However, wehave designed a mechanicalwing which is simply an 80cm. elastic rod connected to avariable potentiometerproviding a varying voltagewhich can then be convertedto digital value through ananalogue to digital converter.Tn this way, we can have acheap sensor which gives acontinuous signal monitoringdistance from the crop edge(and other obstacles). Thesensor configuration for cropharvesting implemented onthe electrical vehicle is shownin Fig. 2a and the diesel robotis shown in Fig. 2b, theoutdoor robots are alsoequipped with ultrasoundsensors wh ich arecharacterised by high noiseimmunity level.

3. The fuzzy logiccontroller designLotfi A. Zadeh introduced thesubject of fuzzy sets in 1965(Zadeh, 1965). In that work.Zadeh suggested that one ofthe reasons humans are beuerat control than conventionalcontrollers is that they are ableto make effective decisions onthe basis of impreciselinguistic information. Heproposed fuzzy logic as a wayof improving the performanceof electromechanicalcontrollers by using it to modelthe way in which humansreason with this type of controlinformation. Figure 3 shows

12

Sensors(LF),(LB),{RF),(RB)

MF

Fig. 4 The membership functions (MF) of the inputsensors; LF,left front; LB, left back; RF, right front; RB,right back

Speeds(left speed),(right speed)

MF

Fig. 5 The membership functions (MF) of the indoorrobot output speeds

Outdoor speedsMF

Fig. 6 The output membership functions (MF) of theoutdoor robot speed

the basic configuration of afuzzy logic controller (FLC),which consists of fourprincipal components whichare the fuzzification interface,knowledge base (comprisingknowledge of the applicationdomain and the attendantcontrol goals), decisionmaking logic (which is thekernel of an FLC), anddefuzzification interface.

In the following analysis.we use a singleton fuzzifier,triangular membershipfunctions, product inference,max-product composition,height defuzzification. Thesetechniques are selected due totheir computational simplicity.The equation that maps thesystem input to output is givenby:

M G

L p=l Y pIT a Aip/e I

G~ AI IIL.- p ~ I a Aip

J'. I

where: M is the totalnumber of rules; y is the crispoutput for each rule;aaAiP isthe product of the membershipfunctions of each rule inputs;and G is the total number ofinputs. More informationabout fuzzy logic can be foundin Lee (1990).

The membership functions(MF) of the inputs denoted byleft front (LF) sensor and theleft back (LB) sensor [rightfront (RF) sensor and rightback (RB) sensor in the case ofthe outdoor robots] are shownin Fig. 4. The outputmembership functions shownin Fig. 5 are the left and rightspeeds for the indoor mobilerobot, the robot steering beingperformed by moving atdifferent wheel speeds. Theoutdoor memberships are thesame for the inputs sensors (inspite of using different sensorsfrom the indoor robots). As theoutdoor robots have a steeringmotor, the output membershipfunctions consist of speed inFig. 6 and the steeringparameters in Fig. 7.

Landwards, Summer 2000

Page 4: Developing afuzzy logic controlled agricultural vehiclevictor.callaghan.info/publications/2000_JAE00(Developing...Fuzzy logic excels in dealing with such imprecise sensors and varying

AUTONOMOUS VEHICLES

challenge to therobot because oftheir irregularity andlow sensitivity ofsonar sensors towardthem. In the nextphase, we have triedthe same architecture10 the outdoorenvironments totrack fences and realcrop edges in realfarms. Eachexperiment wasrepeated five timesand each time thepath was recorded to

test the system repeatability and stabilityagainst different weather and groundconditions (such as rain, wind, holes inthe ground, going up and down hill, etc.).Figure lOa shows the robot emulating thecrop cutting operation. Here it continuesgoing inwards to complete the harvestingoperations. The cutting action wassimulated by reducing the size of thefence. Note that the response is smoothespecially when the robot turns. This isdue to the smooth transition between rulesand the smooth interpolation betweendifferent actions which are characteristicsof fuzzy logic. The same experiment wasrepeated but with real bales of hay andgave a very smooth and a repeatable

[Ste~blg1[ i::J ]llmvKM,aumJ1 High J [ ~~': ]

Fig. 7 The output membershipfunctions (MF) of the outdoorrobot steering

The rule base of the indoor controlleris the same for the outdoor robots exceptfor speed and steering aspects. Also theindoor robot was left edge followingwhile in the outdoor robots it will be lightedge following (a peculiarity of the factthe vehicles were built by differentpeople). These rule bases and themembership functions were designedusing human experience but we aredeveloping methods to learn themautomatically using genetic algorithms.

Figures 8 and 9 represent-the controlsurfaces of the indoor and the outdoorrobots. Figure 8 represents the indoorrobot control surface in which the LF andthe LB were plotted against their outputswhich are the left speed (left figure) andthe right speed (right figure). Figure 9represents the control surface of theoutdoor robots in which RF and RB wereplotted against their outputs which are therobot speed (left figure) and the robotsteering (right figure).

4. Experimental resultsThe performance of the architecture hasbeen assessed in two main ways. Firstly,we physically emulated (rather thansimulated) the crop following process. Inthis emulation, we have conductedpractical experiments with the indoorrobots to track the robots paths andreactions to the irregular geometricalshapes forming fences that fake the cropedge. These fences included one formedwith real bales of hay which are real

Landwards, Summer 2000

Fig. 8 The control surface of the indoor robots; LF,left frontsensor; LB, left back sensor

Fig. 9 The control surface of the outdoor robots; RF, right front sensor;RB, right back sensor

13

Page 5: Developing afuzzy logic controlled agricultural vehiclevictor.callaghan.info/publications/2000_JAE00(Developing...Fuzzy logic excels in dealing with such imprecise sensors and varying

AUTONOMOUS VEHICLES

as mown turf, hay stubble and ground thatwas rutted. It was also tested on flat landas well as on slopes, both up and downand across the slope.

The same control architecture wasused in aUrobots only varying the outputCMF) of the robots and slightly varyingthe rule base to cater for the differingsteering and speed characteristics of therobots. We experimented withmechanical wands and ultra soundsensors. In spite of the varying weatherconditions, the systems displayed a verygood response showing the fuzzycontroller can deal with imprecision andnoise.

Figures /] a and 11b show the robotpath of the electrical outdoor robotfollowing an outdoors fence. InFig. 11a,the robot succeeded in following anirregular rectangular metallic fence underdifferent weather conditions (i.e. wind

Fig. 10 (a) The robot emulating the harvesting operation; and (b) the robot following fences formed bybales of hay

Fig. 11 (a) The outdoor electrical robot following an irregular fenceusing ultra sound sensors; and (b) the outdoor robot following anirregular fence using the mechanical wand sensors

response as in Fig. lOb.We then tried the electric robot in a

range of out-door environments. Theseinvolved following many types of cut

edge, such as rough pasture, hay cropsand bedges. The system was also triedunder different weather conditions andunder different ground conditions, such

Fig. 12 (a) The electrical robot following out door irregular tree hedges; and (b) the robot path

14 Landwards, Summer 2000

Page 6: Developing afuzzy logic controlled agricultural vehiclevictor.callaghan.info/publications/2000_JAE00(Developing...Fuzzy logic excels in dealing with such imprecise sensors and varying

AUTONOMOUS VEHICLES

Fig. 13 (a) The diesel robot following real irregular hay crop edgeusing the mechanical wands; and (b) the robot path

and rain) using only two ultra sound (US)sensors. The robot gave repeatable andsmooth path following on the wholefence, as well as turning around corners.Figure llb shows that the robotsucceeded in following the same fence

successfully within a tolerance of 50 mm.Figure J 3a shows the diesel robot

with the mechanical wand sensors in ahay field that has a very discontinuousedge and ill defined corners. The robotgave stable. repeatable and robustresponse as shown in Fig. J3b, and

Fig. 14 (a) The robot starts turning around an irregular hay cropcorner; and (b) the robot after turning smoothly around the corner

using the mechanical wands, the robotagain following tbe fence with highrepeatability and stability and respondingrapidly but smoothly to any changes inthe fence line.

Figure 12a shows the electrical robotin the field following a crop edge whichis characterised by high irregularity (gapsin the edge, plants falling from the edge).The robot was also required to navigateuphill and downhill in a ground full ofruts. It used two ultra sound sensors tosense the crop edge. Again the robot gavea smooth response and followed the cropkeeping a safe distance from the cropedge and responding rapidly but smoothlyto any changes in the edge Fig. 12b.Although we currently have noquantitative means for evaluating theprecision of the crop following, weestimate that the crop edge was tracked

Landwards, Summer 2000

tracked the edge of the crop successfull ywithin a tolerance of 50 mrn. The robotalso turned smoothly around the ill-defined hay crop corners, as shown in Fig.14£1. Figure 14b shows the robot afterturning smoothly around this comer.

s. ConclusionsIn this paper, we have developed a fuzzycontroller for a robot aimed at automatingthe crop following processes. We havedeveloped a novel sensor design (outdoormechanical wands) to be used in realfarms under different conditions. Wetested the fuzzy control architecture onan indoor mobile robot with only twoultrasound sensors. It had succeeded inmaintaining itself at a constant distancefrom the crop in spite of boundaryirregularities and the imprecision in the

ultrasound sensors. After testing thearchitecture successfully indoors, thecontrol architecture was moved to theoutdoor robots. In the field environment,the robots displayed a smooth and fastresponse and were able to track variousedges under different weather and groundconditions.

The outdoor robots tracked irregularcrop edges successfully within a toleranceof 50 rnm. The robot also turned aroundreal crop corners smoothly and gave ahighly repeatable and stable response. Tothe authors' knowledge, the workdescribed in this paper is the only systemwhich has successfully guided a dieseltractor in the outdoor environment,following real crop edges (includingirregular edges which include gaps) andturning around corners with a high degreeof repeatability and following the cropedge with a tolerance of 50 mm. Thesystem is totally autonomous with no pre-specified plans and reacts in real time tothe changing field conditions.

We are currently investigating theperformance of other farm tasks (such asthe collection of bales of hay or fruitboxes). In these, we are going to use afuzzy hierarchical controller to combineseveral behaviours for safe navigationtoward our goals. In this work, we willintegrate a vision system for bales of haydetection and will try to integrate it withthe fuzzy system for reactive navigation.Also, we are currently investigating theuse of GA based methods in respect toadding a learning capability to thecontroller so that it can adapt itself to thechanging conditions of a field.

ReferencesCallaghan V, Chernett P, Colley M,

Lawson T, Standeven J (1997).Automating agricultural vehicles.Industrial Robot Journal. 5. 346-369.

Cho S, Ki H (1996). Unmanned combineoperation using fuzzy logic controlgraphic simulation. Applied Engineeringin Agriculture, 247-251.

Kosko B (1992). Fuzzy systemsas universalapproximators, Proceedings of the IEEEInternational Conference on FuzzySystems, San-Diego, California. March.pp.1l53-1162.

Lee C (1990). Fuzzy logic in controlsystems: fuzzy logic controller, part I &part II. IEEE Transactions, Systems,Management, Cybernetics, 20. 404-434.

Mandow M et al. (1996). The autonomousmobile robot Aurora for greenhouseoperation. IEEE Robotics and

15

Page 7: Developing afuzzy logic controlled agricultural vehiclevictor.callaghan.info/publications/2000_JAE00(Developing...Fuzzy logic excels in dealing with such imprecise sensors and varying

AUTONOMOUS VEHICLES

Automation Magazine, 3, 18-28.Ollis M, Stentz A (1996). First results in

vision-based crop line tracking.Proceedings of the 1996 IEEEInternational Conference 011Robotics andAutomation, pp. 951-956.

Schallter S (1996). Object recognition inagricultural environment. MScDissertation, Essex University.

Yamasita S (1990). Examination of

induction method of unmanned vehicle forgreenhouse. Proceedings of 49th Congressfor Society of Farm Machinery, Japan,pp. 273-274.

Zadeh LA (1965). Fuzzy sets. InformationConference, Vol. 8, pp. 338-353.

Ziteraya Z, Yamahoso K (1987). Patternrecognition of farm products by linguisticdescription with fuzzy theory.Proceedings of 3'" Fuzzy System

Symposium of JFSA Branch, Japan, pp.127-132.

Zhang Q,Letchfield J,Bentsman J (1990).Fuzzy predictive control systems for cornquality control drying food processingautomation. Proceedings of the 1990Conference, ASAE, St Joseph. Mich., pp.313-320.

Lantr. report points to increasingrequirement for higher skills levelsThe latest research into labour markettrends by Lantra (the National TrainingOrganisation for land-based industries)shows that there is an increasing needfor higher levels of skills in most land-based industries. Traditional unskilledand semi-skilled jobs are in decline andskills levels have to rise to meet thechallenges of new working practices,rapid developments in technology andincreasing competition.

The report, launched at the FarmersClub, London on 19 February states that10 years ago many land-basedemployees needed skills equivalent to.OJ just above, level 2 in a National orScottish Vocational Qualification (N/SVQ). This has increased to level 3 orhigher. Other key findings from thereport include:

workforce turnover is increasing andaverages 14%43% of workers have industry-relevant qualifications at N/SVQlevel 2 or above26% of the workforce is qualifiedto level 3 or abovean average of only 1.3 days per

person was spent on training last yearless than 25% of land-based.bnsinessesarrange formal trainingonly 10% of land-based businesseshave a formal training planThe labour market information report

presents the results of a two-year projectthrough which Lantra has collectedinformation from over 7,500 businessesoperating in all aspects of the Iandbasedsector across Great Britain. As well asproviding the basis for the report, this datais also held in an economic forecastingdatabase which can be used to help predictfuture demand for numbers of workers andtheir skill levels.

In the report's foreword, minister for1ifelong learning Malcolm Wicks said: 'Thedata gathered in this report will serve as asolid foundation on which to develop thenecessary foresight to future skills needsof the land-based industries. Naturally, this.is a continuous process and this report onlyrepresents the first step in a much largerjourney. Lantra must now gather supportfrom employers, education and trainingproviders, trade associations and others todevelop a full Ski lis Foresight report and

to put in place the measures necessary10 address future skills needs.'

Lantra's chairman Andy Stewartsaid: 'The report.lrighlights the difficulttimes now facing manyparts of the land-based sector. These new-challenges willhave. to be met by eXlsti'ng or futurerecruits and are likely to demand newskills and competences. If businessesare to compete effectively, employersmustunderstand which skills are neededand take action to ensure that: they existin the workforce.

The report was used as the basis fora conference on 29 February foremployers, trade association staff andeducation and training providers. 'Thisprovided an opportunity for the sectorto discuss the significance. of Lantra'sfindings and shape a full skills-foresightreport which we aim to publish in May'said Mr Stewart.

Contact: Copies of the report areavailable from Lantra Connect on0345 078007.

Gentle new potato planter boostsplanting accuracyKvemeland has launched a brand new tworow potato planter in time for this Spring'splanting season. Designated the UN3000T, the new planter boasts two majornew benefits aimed at improving plantingefficiency.

The first is an Hydraulically tipping 1.5tonne capacity hopper, which has beenfound to be extremely gentle whenhandling chitted seed. The hopper featureis ideal for larger potato growers whorequire increased efficiency from acompact and manoeuvrable planter,

16

especially when working long runs.The second new feature is electronic

vibrating agitation. The cupped beltvibrates from top to bottom, ensuring thatonly one potato per cup is delivered down.The degree of agitation is variable to suitdifferent working conditions and tubersizes, and the system has proven itself farmore reliable and accurate than standard'one bump' manual systems. The vibratingagitation system can also be fitted to thestandard Kvemeland UN 3000 planter.

The UN 3000T retains many of the well

proven and popular features of the existingUN 3000 range, including a large plantingunit with 74 rom cups as standard to handlelarge sized tubers, a top drive shaft for cupbelts, and spacing that is adjustable from theside of the planter. A ridging hood is alsoavailable in lieu of the standard ploughs.

Retail price for the new UN 3000Tplanter is from £7,990 complete with thenew agitation system.

Contact: Les Davidson, Potato ProductManager, Kverneland (UK)

landwards, Summer 2000