3 Experimental Method 3.1 Spray Uniformity The manipulator is controlled so that the spray nozzle can move below the trellis keeping the distance between nozzle and trellis constant bas
Trang 1Spraying Robot for Grape Production 543
Fig.5 Crawler type traveling device.
3 Experimental Method
3.1 Spray Uniformity
The manipulator is controlled so that the spray nozzle can move below the trellis keeping the distance between nozzle and trellis constant based on distance information from the ultrasonic sensor in order to uniformly spray the target To evaluate the spray results comparing with human, black colored water was used in stead of chemical liquid and was sprayed to a flat white paper which was set above the manipulator end in parallel to the ground
Fig.6 Spraying course.
Trang 2544 Y Ogawa et al.
Figure 6 shows a spraying course by robot It was assumed that this robot sprayed at a unit area after a traveling device stopped moving in order to minimize the influence of vibration caused by a traveling device or rugged ground The area
of spraying by this robot was about 1.5 m2during the traveling device stop The spray nozzle was controlled to move in a linear motion with constant velocity keeping constant distance between the nozzle and the object to spray uniformly The spraying was stopped not to overlap-spray by switching an electric valve when the nozzle moved from a line to the next line The spraying operation by human was also conducted in the same way with robot to compare its uniformity of spraying
Texture analysis methods were used for evaluation of spray uniformity by robot and human Images were acquired by a color TV camera and a capture board which was installed in a PC NTSC analog signal was inputted to the caputure board whose pixel number was 256 × 256 with 256 gray levels Green component images were used for the texture analysis In this experiment, three textural features, ASM (angular second moment), IDM (inverse difference moment) and CON (contrast) were used [7] ASM and IDM indicate the measure of homogeneity in a large area and in a local area of image respectively, while CON indicates the measure of difference of grey level in the whole of image [8]
The texture-context information is adequately specified by the matrix of relative frequencies pij with which two neighboring resolution cells separated by distance d occur on image, one with gray tone i and the other with gray tone j as shown in Figure 7
In a case of d =1 and a =0, the three textural features are calculated as follows:
ASM: ∑∑n i=−l n j=−l p l i j
0 0
2 )}
, )(
0 , (
CON: ∑∑n i=−l n j=−l l p + l i − i j j
0 0
2 ) (
) , )(
0 , (
(2)
IDM: ∑∑n=−l =− −
i
l n
j
j i l p j i
0 0
2
) , )(
0 , ( )
Fig.7 Co-occurrence matrix.
Trang 3Spraying Robot for Grape Production 545
3.2 Obstacle Avoidance
Grapevine trellis is usually not flat but fluctuated according to its leaves, stems and bunches growth It is necessary that the manipulator should be controlled along the fluctuated trellis to uniformly spray In this experiment, the manipulator was continuously path-controlled at a constant speed keeping a same distance between end-effector and the trellis (30 cm)
Fig.8 Distance measuring method.
Figure 8 shows distance measurement and control methods Manipulator moves from right to left in this figure in velocity Vh and the ultrasonic sensor detects distance several times during moving for length S/2 (half pitch of length between ultrasonic sensor and spray nozzle) The average of the measured distances is used for vertical moving distance Z(n)
Although many leaves completely covers trellis somewhere, there are many spots where no leaf grows on actual trellis in the field When the ultrasonic sensor met the no leaf spot, the sensor outputted the maximum value of Z(n) and the manipulator was supposed to be abruptly ordered to move toward out of its operational space To avoid this problem, the maximum distance data was not added to the data for average calculation When more than half of the distance data were the maximum, the manipulator was controlled on horizontal movement Since sampling interval of the ultrasonic sensor was constant (60ms), sampling times during moving for distance S/2 were changed according to the manipulator moving velocity Vh; 20 times for Vh50 mm/s, 10 times for 100 mm/s, and 5 times for 200 mm/s In this experiment, the distance between nozzle and object was set to
300 mm
Trang 4546 Y Ogawa et al.
4 Result and Discussion
4.1 Spray Uniformity
Figure 9 shows a set of gray level histograms (Y axial direction in Figure 6) of spaying pattern on white color flat paper The Robot and human sprayed on the condition that the distance between the nozzle and the object was 300 mm and that velocity of the nozzle was 200 mm/s The horizontal axis indicates pixel number of image and the vertical axis indicates 256 gray levels
From this result, it was observed that gray level value of robot spray was apparently more uniform than that of manual spray and that human could not spray constantly even on flat plane It was predicted that robot could follow the fluctuated grapevine trellis so precisely that the difference of spray uniformity would be larger on an actual trellis than this result
Table 2 shows a result of evaluation by the textural features It was observed that ASM and IDM of robot spray images were higher than those of human spray images, while CON of robot was lower than that of human, which implies that robot could sprayed uniformly From the results, it was considered that robot could spray more uniformly than human not only in whole larger area but also in smaller area
Fig.9 Comparision of spray uniformity.
Table.2 Result of Textural analysis.
Trang 5Spraying Robot for Grape Production 547
4.2 Obstacle Avoidance
Figure 11 shows a result of obstacle avoidance when an artificial trellis was used in room (Figure 10) When the manipulator moving speed was slow, the ultrasonic sensor was supposed to detect the distances at many points The manipulator end was, therefore, able to precisely follow the shape of plant on trellis and to uniformly spray Furthermore, the manipulator could avoid a fruit existed on the way However, the manipulator end could not follow an abrupt large irregular shape sometimes in case of 200 mm/s moving speed as shown in Figure 11
Fig.10 A grapevine trellis.
Fig.11 A result of experiment.
This precise spraying makes prediction of chemical residues on agricultural products possible by a simple calculation, if the spray record (type of chemical, spray quantity per unit area) is kept and the spray information is linked to GIS information It is possible that total quantity of chemicals for protect from insect injuries and disease is determined depending on local region, on crop, on season, and on other conditions Chemical spraying operation has been usually done in
Trang 6548 Y Ogawa et al.
every year, even if no vermin or no disease was found, because chemical spray operation is conducted not for extermination but for prevention When a field monitoring system to find insect injury or diseases on early stage is developed by machine vision recognition adding to this precise spraying technology, it is expected that necessary chemicals can be sprayed only at necessary places for protection of environment and ecosystem by establishment of traceability system
5 Conclusion
From these experimental results, it was considered that the precise spraying operation became possible by using of robot and that prediction of chemical residues on agricultural products was also possible because the chemical spraying operation was easily recorded by the robot system If a traceability system including chemical spray operation is established, an inspection system of chemical residue on agricultural products will be earlier realized because chemical quantity in the field and chemical residues are able to be calculated In addition, a monitoring system to early detect insect injuries and diseases of products is desirable for the minimum chemical spraying
References
1 Kondo, N, “Study on Grape Harvestion Robot”, Proc IFAC/ISHS 1 st Workshop on Material and Control Applications in Agriculture and Horticulture, pp 243-246, 1991.
2 Kondo,N et al, “Basic Studies on Robot to Work in Vineyard (Part 1)”, Journal of the Japanese Society of Agricultural Machinery, 55(6), pp.85-94.(in Japanese), 1993.
3 Kondo,N et al, “Basic Studies on Robot to Work in Vineyard (Part 2)”, Journal of the Japanese Society of Agricultural Machinery, 56(1), pp.45-53.(in Japanese), 1994.
4 Monta,M et al, “Basic Studies on Robot to Work in Vineyard (Part 3)”, Journal of the Japanese Society of Agricultural Machinery, 56(2), pp.93-100.(in Japanese), 1994.
5 Monta,M., Kondo,N and Shibano,Y., 1995a Agricultural robot in grape production
system In Proc 1995 IEEE International Conference on Robotics and Automation,
vol.3: 25042509
6 Kondo, N et al, Robotics for Bioproduction systems, American Society of Agricultural
Engineers, Michigan, USA, 1998
7 Haralick,R.M et al, “Textural Features for Image Classification”, IEEE Transactions on Systems, Man, and Cybemetics, Vol SMC-3, No 6, pp 610-621, 1973.
8 Monta,M., Kondo,N., Shibano,Y and Mohri,K., 1995b End-effectors for agricultural robot to work in vineyard, Acta Horticulturae 399: pp 247-254
Trang 7Path Planning for Complete Coverage with
Agricultural Machines
Michel Ta¨ıx1, Philippe Sou`eres1, Helene Frayssinet1, and Lionel Cordesses2 LAAS-CNRS
7 Av du Colonel Roche,
31077 Toulouse Cedex 4,France
{name}@laas.fr
RENAULT Agriculture, R&D
7 Rue Dewoitine,
78141 V´elizy,France
cordesses@renagri.com
Abstract The problem of planning reference trajectories for agricultural machines is
consid-ered A path planning algorithm to perform various kinds of farm-works is described The case
of convex fields is first considered A direction of work being given, the algorithm determines the turning areas and selects a trajectory which guarantees the complete field coverage while minimizing overlapping The method is extended to the case of fields with more complex shape including possibly obstacles Simulations are proposed to illustrate the reasoning
1 Introduction
This paper presents a research work issued from a collaboration between RENAULT Agriculture and the LAAS-CNRS which concerns the automatic guidance of high-end farm tractors on the base of GPS data Steering strategies can be divided into two classes: relative guidance and absolute guidance Relative guidance consists steering the vehicle by regulating its posture with respect to the track resulting from the previous passage (crop or ploughing line) In that case, trajectories are often rectilinear and parallel Absolute guidance consists in tracking a reference path, or
a trajectory, issued from a path planning strategy [4], [6] Our work deals with the absolute guidance problem It focuses on the description of a trajectory planning algorithm which provides a field coverage strategy adapted to various kinds of farm-works [15], [10] The main difficulty of the problem comes from the need
to realize the complete covering of the field, that is including the regions inside which the manoeuvre are executed Planning the trajectories inside the manoeuvre area states a difficult problem which is crucial for agricultural applications Indeed, while these zones are usually covered at the end when ploughing, they need to be worked at the beginning when harvesting Previous work devoted to the coverage problem only provide algorithms for the case of simple rectangular areas and do not address the planning problem inside manoeuvre areas In [16], [1] [5], [13], cellular decomposition approaches have been proposed based on breaking down the workspace The Spiral-STC algorithm proposed in [9] is based on a discretization of the working area and the definition of a spanning tree to solve coverage Considering fields with more complex shapes states another difficult problem Indeed, in that case the working direction may differ from a region to another and a cell decomposition
S Yuta et al (Eds.): Field and Service Robotics, STAR 24, pp 549–558, 2006.
© Springer-Verlag Berlin Heidelberg 2006
Trang 8550 M Ta¨ıx et al.
has to be done Such an approach is proposed in [11] where a sequence of sub-regions is selected with different planar sweep lines to compute the coverage path Theoretical results based on computational geometry can be found in [2], [3] The algorithm presented in this paper allows to determine automatically the manoeuvre areas and select a covering trajectory which minimizes overlapping The planning approach is first presented in the case of convex fields Two strategies are proposed to this end On this base the presence of obstacle is then considered and the method is extended to the case of fields with more convex shape
2 The Automatic Guidance Project
The path planning algorithm presented in this paper comes as a part of an industrial project of RENAULT Agriculture which aims at developing autonomous navigation abilities for farm tractors
Control Law
GPS
δδ
Reference path ++
−−
εε
εε : error
δδ : steering angle
Fig 1 Farm tractor control system overview
The GPS-based farm tractor control system is based upon the following four units (figure 1):
• The sensor: Real-time, kinematic GPS Its high three dimensional (3D) accuracy (σ < 2cm) and its low latency (tlatency< 0.2s, see [7]) allow its use in a closed loop system It outputs information about position and velocity of one point of the vehicle to control
• The farm tractor to control: The only technical requirement is the availability
of a model with an electro-hydraulic power steering instead of an all-hydraulic one The steering angle can by supplied either by the driver, thanks to the driving wheel, or by the embedded computer
• The Controller implemented on an embedded computer: The system is able to follow paths at various velocities [14,6] with an accuracy better than 10cm
• The trajectory planner which determines the reference path to follow to perform
a specific farm-work
This paper focuses on the fourth unit only, namely the path planning problem
3 Covering Path Planning
Farm-work experiments have proven that the choice of the working direction within the field has to be guided by two major factors First, to reduce sliding and traction
Trang 9Path Planning for Complete Coverage with Agricultural Machines 551 efforts, the tractor must move at best in the direction of the slope and execute trajectories with very low curvature Second, to reduce the number of manoeuvres, the direction of motion must be, as far as possible, parallel to the longer side of the field In particular, in wedge-shaped regions, the motion must be parallel to one of the edges To satisfy these constraints at best, it appears necessary to decompose the
field into regions, and define in each of them a “Steering edge” S-edge which will
guide the successive tracks Furthermore, when planning trajectories, it is necessary
to determine regions called “Turning areas” T-areas, located at extremities of the
field, inside which the tractor will execute U-turns or manoeuvres The width of T-areas depends on the tractor’s characteristics and the nature of the tool
Fig 2 Definitions
The remaining part of the field constitutes the “working-area”, W-area Inside
this central region, the farm-work trajectories are most part of time rectilinear parallel
tracks directed along the S-edge.
The algorithm proposed in this paper applies to polygonal fields including at most one vertex of concavity An extension is proposed to consider the case of fields including one moderate curved boundary, that is one smooth low-curved boundary along which the tractor can move This restriction allows to consider most part
of fields encountered in real applications For such a field, once the input area,
I-area, and the output I-area, O-I-area, have been defined on the field’s boundary, the
path-planning problem can be stated as follows:
Determine a trajectory starting from a point in the I-area and ending at a point of the O-area which guarantees the coverage of the whole field (W-area + T-areas) while minimizing the overlapping between adjacent tracks and the number
of manoeuvres.
Note that, depending on the nature of farm-work, the covering of the T-areas is done at the beginning or at the end of the task For instance, when ploughing the
T-areas are to be covered at the end, while they are worked at the beginning during
harvest The algorithm is based on the partitioning of the field into convex polygons The partitioning process is described in section 3.3 Inside each convex polygon, a
S-edge is determined and a set of characteristic points is defined at the boundary of
the W-areas and the T-areas These points will constitute the nodes of a graph upon which the trajectory is defined Two strategies are proposed to this end The trajectory planning strategy is first described for the case of a convex polygon free of obstacles
in section 3.1 The presence of obstacles is considered in section 3.2 Depending on
Trang 10552 M Ta¨ıx et al.
the size of the obstacles two avoidance strategies are proposed Finally, section 3.4 describes the extension of the method to the case of fields including one moderate curved border
3.1 Case of Convex Polygonal Fields
This section presents the trajectory planning method for the case of a convex polygon
free of obstacles The input data are the S-edge, the I-area, the O-area, the kind of
farm-work to be executed and the characteristics of the tractor and the tool (type, width, curvature radius)
The path planning strategy is based on three successive steps The first one
is a topological representation of the field which consists of determining a set of characteristic points from which a graph is defined (section 3.1) On this base, two strategies are proposed to construct the reference trajectory
Determination of characteristic points Once the S-edge is specified, the T-areas
are computed by taking into account the space required to perform the turning manoeuvres In practice, this space is a whole number of the tracks width This implies to shift or add a pair of characteristic points to guarantee the field coverage without overflow Outside the T-areas, the field is covered by parallel tracks directed along the S-edge The tracks are arranged in such a way to insure the complete field covering while minimizing overlapping Following the same technique, the T-areas are also covered by parallel tracks but directed along the side-edges The end points
of all working-tracks are considered as characteristic points (see figure 3 left)
Construction of the trajectory In order to construct of the trajectory, the
charac-teristic points are considered as the nodes of a graph Two strategies are proposed
to define the arcs and explore this graph The first one is based on the search for the best Hamiltonian path according to the minimization of a cost criterion, while the second involves a simpler geometric reasoning
charac-teristic points defined by the end points of tracks These points are considered as the nodes of a graph A set of graph edges U = {u1, u2, , um} is then defined, representing rectilinear paths between these nodes from which the different kind of farm-work can be synthesized To achieve a given farm-work, a specific value is assigned to the graph edges The coverage strategy is deduced from a search within this graph G(X, U) Seven types of edges are to be considered depending on the kind of displacement they represent A specific value piis associated to each type (see figure 3 right):
p1: to execute a working track inside the field,
p2: to pass from a working track to the next one,
p3: to jump from a working track to a track located one after the next one (to avoid manoeuvres),
p4: to jump from a working track to any other track except the next two,
p5: to jump from a working track to a point located inside the T-area,