Certains contenus de cette application ne sont pas disponibles pour le moment.
Si cette situation persiste, veuillez nous contacter àObservations et contact
1. (WO2018087546) SYSTÈME ROBOTISÉ DE CUEILLETTE DE FRUITS
Note: Texte fondé sur des processus automatiques de reconnaissance optique de caractères. Seule la version PDF a une valeur juridique

CLAIMS

1. A robotic fruit picking system comprising an autonomous robot that includes the following subsystems:

a positioning subsystem operable to enable autonomous positioning of the robot using a computer implemented guidance system, such as a computer vision guidance system;

at least one picking arm;

at least one picking head or other type of end effector, mounted on each picking arm to either cut a stem or branch for a specific fruit or bunch of fruits or pluck that fruit or bunch, and then transfer the fruit or bunch;

a computer vision subsystem to analyse images of the fruit to be picked or stored;

a control subsystem that is programmed with or learns picking strategies;

a quality control (QC) subsystem to monitor the quality of fruit that has been picked or could be picked and grade that fruit according to size and/ or quality; and

a storage subsystem for receiving picked fruit and storing that fruit in containers for storage or transportation, or in punnets for retail.

2. The robotic fruit picking system of Claim 1 in which the system further comprises a tracked or wheeled rover or vehicle capable of navigating autonomously using a computer vision-based guidance system.

3. The robotic fruit picking system of any preceding Claim in which the computer vision subsystem comprises at least one 3D stereo camera.

4. The robotic fruit picking system of any preceding Claim in which the computer vision subsystem that analyses fruit images comprises image processing software for detecting a fruit, and the control subsystem comprises software for deciding whether to pick the fruit and the optimal strategy for picking the fruit, based on automatically updateable strategies, such as reinforcement learning based strategies.

5. The robotic fruit picking system of any preceding Claim in which the control subsystem automatically learns fruit picking strategies using reinforcement learning.

6. The robotic fruit picking system of any preceding Claim in which the picking arm has 6 degrees-of-freedom.

7. The robotic fruit picking system of any preceding Claim in which the picking arm positions the end effector and a camera, each mounted on the picking arm.

8. The robotic fruit picking system of any preceding Claim in which the end effector comprises a means of (i) cutting the fruit stalk or stem and (ii) gripping the cut stalk or stem to transfer the fruit to the QC and storage subsystems.

9. The robotic fruit picking system of any preceding Claim in which the robot automatically loads and unloads itself onto, and off of, a storage container or a transport vehicle.

10. The robotic fruit picking system of any preceding Claim in which the robot automatically navigates amongst fruit producing plants, such as along rows of apple trees or strawberry plants, including table grown strawberry plants, or raspberry plants.

11. The robotic fruit picking system of any preceding Claim in which the system automatically collaborates with other robotic systems and human pickers to divide picking work efficiently.

12. The robotic fruit picking system of any preceding Claim in which the system automatically determines the position, orientation, and shape of a target fruit.

13. The robotic fruit picking system of any preceding Claim in which the system automatically determines whether a fruit is suitable for picking based on factors which are automatically updateable in the quality control subsystem.

14. The robotic fruit picking system of any preceding Claim in which the end effector separates the edible and palatable part of a ripe fruit from its stem or stalk without contacting the edible part.

15. The robotic fruit picking system of any preceding Claim in which the system automatically grades a fruit by size and other measures of suitability that are programmed in to, or learnt by, the QC subsystem.

16. The robotic fruit picking system of any preceding Claim in which the system automatically transfers a picked fruit to a suitable storage container held in the storage subsystem without handling the edible and palatable part of the fruit or other sensitive parts of the fruit that could be bruised by handling.

17. The robotic fruit picking system of any preceding Claim in which the control subsystem minimises the risk of the end effector or other part of the robot damaging a fruit or plant on which the fruit grows using machine learning based picking strategies.

18. The robotic fruit picking system of any preceding Claim in which the picking arm moves an attached camera to allow the computer vision subsystem to locate target fruits and determine their pose and suitability for picking.

19. The robotic fruit picking system of any preceding Claim in which the picking arm is a light weight robotic arm with at least some joints that exhibit a range of motion of +/- 275 degrees that positions the end effector for picking and moves picked fruit to the QC subsystem.

20. The robotic fruit picking system of any preceding Claim in which the control subsystem operates the total positioning system and the picking arm.

21. The robotic fruit picking system of any preceding Claim in which the control subsystem uses input from the computer vision subsystem that analyses fruit images to decide where and when to move the robot.

22. The robotic fruit picking system of any preceding Claim in which the QC subsystem is responsible for grading picked fruit, determining its suitability for retail or other use, and discarding unusable fruit.

23. The robotic fruit picking system of any preceding Claim in which the robot picks rotten or otherwise unsuitable fruit and then discards that fruit into a suitable container within the robot or onto the ground, and that container is accessible via a discard chute with its aperture positioned at the bottom of the QC subsystem so that the picking arm can drop the fruit immediately without the need to move to an alternative container.

24. The robotic fruit picking system of any preceding Claim in which positive or negative air pressure is induced in a discard chute or an imaging chamber to ensure that fungal spores coming from previously discarded fruit are kept away from healthy fruit in the imaging chamber.

25. The robotic fruit picking system of any preceding Claim in which the system comprises one or more 6-axis light weight robotic picking arms with some or all joints that exhibit a range of motion of +/- 275 degrees.

26. The robotic fruit picking system of any preceding Claim in which the system comprises two or more picking arms and the picking arms are positioned asymmetrically on the robot.

27. The robotic fruit picking system of any preceding Claim in which the robot has tracks that are removable and the robot can run on rails if the tracks are removed.

28. The robotic fruit picking system of any preceding Claim in which the robot is equipped with fruit holding trays that are suspension mounted.

29. The robotic fruit picking system of any preceding Claim in which the robot is equipped with fruit holding trays that are mounted on movable arms that move from a first extended position to a second, more compact position.

30. The robotic fruit picking system of any preceding Claim in which the robot is equipped with fruit holding trays arranged as two or more vertically oriented stacks.

31. The robotic fruit picking system of any preceding Claim in which the robot is powered from a remote power source.

32. The robotic fruit picking system of any preceding Claim in which the robot has one or more lights that activate when a fruit tray or holder needs to be replaced.

33. The robotic fruit picking system of any preceding Claim in which a fast moving robot removes trays or holders automatically from a slower moving robot that does the fruit picking.

34. The robotic fruit picking system of any preceding Claim in which the robot has one or more lights that activate in response to a user input and shine an identifying signal above the robot.

35. The robotic fruit picking system of any preceding Claim in which the system includes an imaging or analysis chamber in which fruit is placed by the picking arm and is then imaged or analysed for grading or quality control purposes.

36. The robotic fruit picking system of any preceding Claim in which the system includes an imaging or analysis chamber in which fruit is imaged or analysed for grading or quality control purposes and in which the imaging or analysis chamber includes an aperture and a chimney or cylinder or lid or baffle on top of the imaging or analysis chamber's aperture that is designed to block unwanted light from entering the chamber, whilst still permitting fruit to be lowered or passed into it.

37. The robotic fruit picking system of any preceding Claim in which the system includes an imaging or analysis chamber in which fruit is imaged or analysed for grading or quality control purposes and in which the imaging or analysis chamber includes one or more cameras and/or other sensors, such as cameras sensitive to specific (and possibly non-visible) portions of the EM spectrum including IR, (ii) cameras and illuminators that use polarised light, and (iii) sensors specific to particular chemical compounds that might be emitted by the fruit.)

38. The robotic fruit picking system of any preceding Claim in which the system includes a cable management system for cables that run through the articulating joints of a robot, the cable management system including a cable enclosure and a central cable

guide that twists relative to the enclosure, allowing a coil or spiral of cable to expand and contract as the joints rotate.

39. The robotic fruit picking system of claim 38 in which the robot includes a picking arm made up of several individual rigid bodies, each attached to another rigid body at an articulating joint, and there is a cable enclosure associated with one or more of each of the articulating joints."

40. The robotic fruit picking system of Claim 38 - 39 in which the cable guide is configured such that the cable is ducted away through the centre of the cable enclosure to the next body.

41. The robotic fruit picking system of Claim 38 - 40 in which the cable management system is configured to minimise the change of local curvature of the cable as the articulating joints move through their full range of motion.

42. The robotic fruit picking system of Claim 38 - 41 in which the cables are unscreened and the enclosure provides screening.

43. The robotic fruit picking system of Claim 38 - 42 in which the cables also serve to provide sufficient heat to reduce cable degradation.

44. The robotic fruit picking system of any preceding Claim in which the picking arm is adjustable and can be repositioned to maximize picking efficiency for a particular crop variety or growing system, such as for the height of a specific table top growing system.

45. The robotic fruit picking system of any preceding Claim in which the system is configured to perform several functions in addition to picking, including the ability to spray weeds or pests with suitable herbicides and pesticides, or to reposition or prune trusses to facilitate vigorous fruit growth or subsequent picking.

46. The robotic fruit picking system of any preceding Claim in which the robot estimates its position and orientation with respect to a crop row by measuring its position and/ or orientation relative to a tensioned cable.

47. The robotic fruit picking system of any preceding Claim in which the robot estimates its position and orientation with respect to a crop row by measuring its displacement relative to a tensioned cable that runs along the row (a vector cable').

48. The robotic fruit picking system of any preceding Claim in which the robot includes one or more follower arms that are mounted to follow the robot.

49. The robotic fruit picking system of Claim 48 in which the follower arm is connected at one end to the robot chassis by means of a hinged joint and at the other to a truck that runs along the cable.

50. The robotic fruit picking system of Claim 49 in which the angle at the hinged joint is measured to determine the displacement relative to the cable.

51. The robotic fruit picking system of Claim 49 - 50 in which the angle is measured from the resistance of a potentiometer.

52. The robotic fruit picking system of Claim 48 - 51 in which two follower arms are used to determine displacement and orientation relative to the vector cable.

53. The robotic fruit picking system of Claim 46 - 52 in which a computer vision guidance system measures the displacement of the robot relative to the vector cable.

54. The robotic fruit picking system of Claim 47 - 53 in which the computer vision system measures the projected position of the cable in 2D images obtained by a camera mounted with known position and orientation in the robot coordinate system.

55. The robotic fruit picking system of Claim 47 - 54 in which a bracket allows the vector cable to be attached to the legs of tables on which crops grow.

56. The robotic fruit picking system of Claim 49 - 55 in which the truck is equipped with a microswitch positioned so as to break an electrical circuit with the truck loses contact with the cable.

57. The robotic fruit picking system of Claim 48 - 56 in which magnetic coupling is used to attach an outer portion of the follower arm to an inner portion of the follower arm, such that in the event of a failure or other event the portions of the follower arm can separate without damage.

58. The robotic fruit picking system of Claim 48-57 in which the separation of an outer portion and an inner portion of the follower arm triggers a control software to stop the robot.

59. The robotic fruit picking system of any preceding Claim in which the picking arm is controlled to optimize trade-off between picking speed and picking accuracy.

60. The robotic fruit picking system of any preceding Claim in which the control sub-system determines the suitability of a specific target fruit or bunch for picking via a particular approach trajectory by determining the statistical probability that an attempt to pick that target will be successful.

61. The robotic fruit picking system of Claim 60 in which estimating the probability of picking success is determined from images of the scene obtained from viewpoints near a particular target fruit.

62. The robotic fruit picking system of Claim 60 - 61 in which determining the statistical probability is based on a multivariate statistical mode, such as monte carlo simulation.

63. The robotic fruit picking system of Claim 62 in which the statistical model is trained and updated from picking success data obtained by working robots.

64. The robotic fruit picking system of any preceding Claim in which the control subsystem determines the probability of collision between a picking arm and an object using an implicit 3D model of the scene formed by the range of viewpoints from which a target fruit can be observed without occlusion.

65. The robotic fruit picking system of any preceding Claim in which the control sub-system determines one or more viewpoints from which the target fruit appears un-occluded, and hence identifies an obstacle free region of space.

66. The robotic fruit picking system of any preceding Claim in which the computer vision subsystem uses a statistical prior to obtain a maximum likelihood estimate of the values of the shape parameters of a target fruit and the system then calculates an estimate of the volume of that target, and from that estimates the weight of the target.

67. The robotic fruit picking system of any preceding Claim in which the computer vision subsystem determines the fruit's size and shape.

68. The robotic fruit picking system of any preceding Claim in which the computer vision subsystem determines the fruit's size and shape as a means of estimating the fruit's and thereby of ensuring that the require mass of fruit is placed in each punnet according to the requirements of the intended customer for average or minimum mass per punnet.

69. The robotic fruit picking system of any preceding Claim in which picked fruit is automatically allocated into specific punnets (or containers) based on size and quality measures of the picked fruit to minimize the statistical expectation of total cost according to a metric of maximizing the expected profitability for a grower.

70. The robotic fruit picking system of any preceding Claim in which a probability distribution describing the size of picked fruits and other measures of quality is updated dynamically as fruit is picked.

71. The robotic fruit picking system of any preceding Claim in which the picking armplaces larger strawberries in punnets that are more distant from the base of the picking arm, in order to minimize the number of time-consuming arm moves to distant punnets.

72. The robotic fruit picking system of any preceding Claim in which the picking arm places selected fruits in a separate storage container for subsequent scrutiny and re- packing by a human operator, if the quality control subsystem identifies those selected fruits as requiring scrutiny by a human operator.

73. The robotic fruit picking system of any preceding Claim in which the control subsystem implements a two-phase picking procedure for bunches of fruit, in which first the entire bunch is picked and second, unsuitable individual fruits are removed from it.

74. The robotic fruit picking system of any preceding Claim in which the robot measures the pose of the picked fruit, so that the fruit can be positioned at the optimal pose for imaging or analysis or for release at the optimal height to fall into a punnet or container.

75. The robotic fruit picking system of any preceding Claim in which the robot determines the position of other picked fruit already in a punnet or container and varies the release position or height into the punnet or container accordingly for new fruit to be added to the punnet or container.

76. The robotic fruit picking system of any preceding Claim in which the robot automatically positions or orients picked fruit in a punnet or other container to maximize visual appeal.

77. The robotic fruit picking system of any preceding Claim in which the robot automatically generates a record of the quality or other properties of a fruit in a specific punnet and adds a machine readable image to that punnet that is linked back to that record.

78. The robotic fruit picking system of any preceding Claim in which the robot chooses paths within free regions of the ground so as to distribute routes over the surface of the ground in a way the optimizes the trade off between journey time and damage to the ground.

79. The robotic fruit picking system of any preceding Claim in which a chain of several robots automatically follow a single "lead' robot that is driven under human control.

80. The robotic fruit picking system of any preceding Claim in which information about the position of several robots and the urgency of any fault condition, or impending fault condition, affecting one or more robots is used to plan a human supervisor's route amongst them.

81. The robotic fruit picking system of any preceding Claim in which the position of the robot with respect to target fruit is controlled to optimize picking performance, such as minimizing expected picking time.

82. The robotic fruit picking system of any preceding Claim in which collision-free paths or obstacle free trajectories are identified in advance of run-time by physical simulation of the motion of the robot between one or more pairs of points in the configuration space, thereby defining a graph (or 'route map') in which the nodes correspond to configurations (and associated end effector poses) and edges correspond to valid routes between configurations.

83. The robotic fruit picking system of any preceding Claim in which a robot arm path planning is built by mapping between regions of space ( voxels') and edges of a route map graph corresponding to configuration space paths that would cause the robot to intersect that region during some or all of its motion.

84. The robotic fruit picking system of any preceding Claim in which the system logs undesirable conditions in the environment that might require subsequent human intervention along with a map coordinate.

85. The robotic fruit picking system of any preceding Claim in which the system stores locations of all detected fruit (whether ripe or unripe) in computer memory in order to generate a yield map.

86. The robotic fruit picking system of Claim 85 which the yield map enables a farmer to identify problems such as disease or under- or over-watering.

87. The robotic fruit picking system of any preceding Claim in which the system stores a map coordinate system position of unripe fruits that have been detected but not picked in computer memory.

88. The robotic fruit picking system of Claim 85 - 87 in which the yield map takes into account the impact on time on the ripeness of previously unripe fruits.

89. The robotic fruit picking system of any preceding Claim in which the system measures the degree to which the robot is leaning over and compensates for the degree of lean by adapting models of the scene's geometry and camera viewpoints accordingly.

90. The robotic fruit picking system of any preceding Claim in which the system includes an accelerometer.

91. The robotic fruit picking system of Claim 89 - 90 in which the degree of lean is directly measured using the accelerometer or indirectly by measuring the position of a part of the robot in a coordinate frame based on the crop row.

92. The robotic fruit picking system of Claim 89 - 91 in which an appropriate 3D to 3D transformation required to correct for the lean is applied to pre-defined camera poses and environment geometry.

93. The robotic fruit picking system of Claim 89 - 92 in which the lateral position of the robot's tracks in the row is dynamically adjusted so that the picking arms are closer to their design position despite the lean.

94. The robotic fruit picking system of any preceding Claim in which oscillations of a fruit caused by picking the fruit are reduced by damping achieved by a soft gripper.

95. The robotic fruit picking system of any preceding Claim in which in which oscillations of a fruit caused by picking the fruit are reduced by damping achieved by modulating the acceleration or velocity or movement of the robot arm's end-effector.

96. The robotic fruit picking system of any preceding Claim in which the system estimates the mass and pendulum length of the fruit.

97. The robotic fruit picking system of Claim 94 - 96 in which the system designs a deceleration or acceleration profile (dynamically or otherwise) required to minimize the amplitude or duration of oscillations.

98. The robotic fruit picking system of any preceding Claim in which the end effector uses at least the following phases:

(i) a selection phase during which the target fruit is physically partitioned or separated from the plant or tree and/or other fruits growing on the plant/ tree and/or growing infrastructure and

(ii) a severing phase during which the target fruit is permanently severed from the plant/ tree.

99. The robotic fruit picking system of any preceding Claim in which the end effector includes a hook.

100. The robotic fruit picking system of any preceding Claim in which the fruit is moved away from its original growing position during the selection phase.

101. The robotic fruit picking system of Claim 98 - 100 in which a decision phase is introduced after the selection phase and before the severing phase.

102. The robotic fruit picking system of Claim 101 in which the decision phase includes rotation of the fruit by its stem or otherwise.

103. The robotic fruit picking system of Claim 101 in which the decision phase is used to determine whether or not to sever the fruit, or the manner in which the fruit should be severed.

104. The robotic fruit picking system of Claim 98 - 103 in which the selection phase is made reversible.

105. The robotic fruit picking system of Claim 104 in which the reversibility is accomplished by a change of shape of the hook.

106. The robotic fruit picking system of Claim 104 in which the reversibility is accomplished by movement or rotation of the hook.

107. The robotic fruit picking system of any preceding Claim in which the system is capable of simultaneously gripping and cutting the stem of a target fruit.

108. The robotic fruit picking system of any preceding Claim in which the system includes multiple picking units located on a single multiplexed end effector.

109. The robotic fruit picking system of Claim 108 in which multiple picking functions on a picking unit are driven off a single actuator or motor, selectively engaged by lightweight means, such as: electromagnets, an engaging pin, rotary tab, or similar.

110. The robotic fruit picking system of Claim 108 - 109 in which a single motor or actuator drives one function across all units on the head, selectively engaged by means such as: an electromagnet, an engaging pin, rotary tab, or similar.

111. The robotic fruit picking system of Claim 109 - 110 in which the functions are driven by lightweight means from elsewhere in the system, such as using: a bowden cable, torsional drive cable/spring, pneumatic or hydraulic means.

112. The robotic fruit picking system of any preceding Claim in which the end effector pulls the target fruit away from the plant in order to determine the fruit's suitability for picking before the fruit is permanently severed from the plant.

113. The robotic fruit picking system of any preceding Claim in which the end effector comprises a hook with a dynamically programmable trajectory.

114. The robotic fruit picking system of any preceding Claim in which the end effector uses at least the following phases:

(i) a selection phase during which the target fruit is physically partitioned or separated from the tree and/or other fruits growing on the tree and/or growing infrastructure;

(ii) a severing phase during which the target fruit is permanently severed from the tree; and

in which the selection and severing phases are performed by the actuation of a loop, wherein the loop diameter, position and orientation are programmatically controlled.

115. The robotic fruit picking system of any preceding Claim in which the end effector uses at least the following phases:

(i) a selection phase during which the target fruit is physically partitioned or separated from the tree and/or other fruits growing on the tree and/or growing infrastructure;

(ii) a severing phase during which the target fruit is permanently severed from the tree; and

and in which the selection and severing phases are performed by a set of jaws, wherein the jaws diameter and position are programmatically controlled.

116. The robotic fruit picking system of Claim 115 in which the jaw attitude such as opened, partially closed or closed is programmatically controlled.

117. The robotic fruit picking system of any preceding Claim in which the computer vision based system is used in order to determine heading and lateral positions of the robot with respect to a row of crops using images obtained by a forwards or backwards facing camera pointing approximately along the row.

118. The robotic fruit picking system of any preceding Claim in which the computer vision based subsystem detects a target fruit and in which the robot includes an end effector wherein part of the end effector is used as an exposure control target.

119. The robotic fruit picking system of any preceding Claim in which the control system software uses lighting conditions inferred or derived from the weather forecast as an input to the control subsystem or computer vision subsystem to control picking strategies or operations.

120. The robotic fruit picking system of any preceding Claim in which the computer vision based subsystem detects a target fruit and a end effector is able to physically separate a candidate fruit further from the plant and other fruits in the bunch before picking.

121. The robotic fruit picking system of any preceding Claim in which mirrors are positioned and oriented to provide multiple virtual views of the fruit.

122. The robotic fruit picking system of any preceding Claim in which the computer vision based system obtains multiple images of a target fruit under different lighting conditions and infers information about the shape of the target fruit.

123. The robotic fruit picking system of any preceding Claim in which the computer vision based subsystem uses an image segmentation technique to provide an indication of a fruit health.

124. The robotic fruit picking system of any preceding Claim in which the computer vision based subsystem detects the positions or points of the fruit achenes or drupelets and assigns a cost to those positions or the arrangement of those positions using an energy function that assigns a lowest energy to regularly arranged positions.

125. The robotic fruit picking system of any preceding Claim in which a semantic labelling approach is used to detect achenes, such as a decision forest classifier.

126. The robotic fruit picking system of Claim 124 - 125 in which the sum of costs over points provides an indication of fruit health.

127. The robotic fruit picking system of any preceding Claim in which an indication of fruit health is provided from analysing one or more of the following: colour of the achenes, colour of the flesh of the fruit or 3D shape of the fruit.

128. The robotic fruit picking system of Claim 127 in which a neural network or other machine learning system, trained from a database of existing images with associated expert-derived ground truth labels, is used.

129. The robotic fruit picking system of any preceding Claim in which the computer vision based subsystem is used to classify a fruit, and in which the system allows a grower to adjust thresholds for classifying the fruit.

130. The robotic fruit picking system of any preceding Claim in which the computer vision based subsystem locates specific parts of a plant or specific plants that require a targeted localized application of chemicals such as herbicides or pesticides.

131. The robotic fruit picking system of any preceding Claim in which the computer vision based subsystem detects instances of specific kinds of pathogen such as: insects, dry rot, wet rot.

132. The robotic fruit picking system of any preceding Claim in which the computer vision subsystem detects drupelets or achenes using specularities induced on the surface of a fruit by a single point light source.

133. The robotic fruit picking system of Claim 132 in which an end effector is used for picking and another end effector is used for spraying.

134. The robotic fruit picking system of any preceding Claim in which an end effector is a spraying end effector that contains a small reservoir of liquid chemical.

135. The robotic fruit picking system of any preceding Claim in which the picking arm visits a station on the chassis to gather a cartridge of chemicals.

136. The robotic fruit picking system of any preceding Claim in which the picking arm visits a cartridge in the chassis to suck the required liquid chemicals from a cartridge into its reservoir or to expel unused chemical from its reservoir back into a cartridge.

137. The robotic fruit picking system of Claim 130-136 in which several different types of chemical are combined in dynamically programmable combination to achieve more optimal local treatment.

138. The robotic fruit picking system of Claim 131 - 136 in which multiple cartridges contain multiple different chemical combinations.

139. The robotic fruit picking system of any preceding Claim in which a machine learning approach is used to train a detection algorithm to automatically detect a target fruit.

140. The robotic fruit picking system of any preceding Claim in which the system identifies fruit in RGB color images obtained by a camera.

141. The robotic fruit picking system of any preceding Claim in which the system identifies fruit in depth images obtained by dense stereo or otherwise.

142. The robotic fruit picking system of Claim 139 - 141 in which the training data is a dataset in which the position and orientation of target fruit is annotated by hand in images of plants that are representative of those likely to be obtained by the camera.

143. The robotic fruit picking system of any preceding Claim in which a detection algorithm is trained to perform semantic segmentation on the images captured by the camera.

144. The robotic fruit picking system of Claim 143 in which the semantic segmentation labels each image pixels such as ripe fruit, unripe fruit or other object.

145. The robotic fruit picking system of Claim 144 in which a clustering algorithm aggregates the results of the semantic segmentation.

146. The robotic fruit picking system of Claim 139 - 145 in which the machine learning approach is a decision forest classifier.

147. The robotic fruit picking system of Claim 139 - 146 in which the machine learning approach is a convolutional neural network.

148. The robotic fruit picking system of Claim 139 - 147 in which a convolutional neural network is trained to distinguish image patches containing a target fruit at their center from image patches that do not.

149. The robotic fruit picking system of Claim 139 - 148 in which a sliding window approach is used to determine the positions of all images likely to contain a target fruit.

150. The robotic fruit picking system of Claim 139 - 149 in which a semantic segmentation is used to identify the likely image locations of a target fruit for subsequent more accurate classification or pose determination by a CNN or other form of inference engine.

151. The robotic fruit picking system of any preceding Claim in which a machine learning approach with a regression model is used to predict the angles describing the orientation of approximately rotationally symmetric fruit from images, including monocular, stereo and depth images.

152. The robotic fruit picking system of any preceding Claim in which a machine learning approach is used to train a detection algorithm to identify and delineate stalks in images captured by a camera.

153. The robotic fruit picking system of any preceding Claim in which a machine learning approach is used to train a prediction algorithm to predict how much improvement to an initial pose estimate for a target fruit is likely to be revealed by a given additional viewpoint.

154. The robotic fruit picking system of any preceding Claim in which the system predicts which additional information, including which viewpoints out of a set of available viewpoints, is likely to be the most valuable, including the most beneficial to overall productivity.

155. The robotic fruit picking system of Claim 154 in which the additional information is the location or point where the stalk attaches to the target fruit.

156. The robotic fruit picking system of Claim 154 in which the additional information is the knowledge that the fruit is visible without occlusion from a particular viewpoint.

157. The robotic fruit picking system of any preceding Claim in which the additional information is the knowledge that the space between the camera and the fruit is free of obstacles from a particular viewpoint.

158. The robotic fruit picking system of any preceding Claim in which the system recovers a 3D shape of a target fruit from one or more images of the target fruit obtained from one or more viewpoints, and in which a generative model of the target fruit's image appearance is used.

159. The robotic fruit picking system of any preceding Claim in which a geometric and/ or photometric model fitting approach is used to predict the surface appearance of a target fruit as well as the shadows cast by the target fruit onto itself under different, controlled lightning conditions.

160. The robotic fruit picking system of Claim 159 in which the cost function, namely the measure of agreement between images, is made robust to occlusion or the fruit is physically separated from sources of occlusion.

161. The robotic fruit picking system of any preceding Claim in which a machine learning approach is used to train a labeling algorithm to automatically assign a label to an image captured by the system, wherein pre-labeled images provided by human experts are used to train the system.

162. The robotic fruit picking system of any preceding Claim in which labelling data provided by human experts is used to train a machine learning system to assign quality automatically to newly picked fruit, by training an image classifier with training data comprising (i) images of the picked fruit obtained by the QC subsystem and (ii) associated quality labels provided by the human expert.

163. The robotic fruit picking system of any preceding Claim in which the control policy subsystem is trained via reinforcement learning while the robot is operating.

164. The robotic fruit picking system of any preceding Claim in which the control subsystem is trained via reinforcement learning, and wherein training is done by simulating the movements of the robot using images of the real-world environment captured amongst available viewpoints.

165. The robotic fruit picking system of any preceding Claim in which the control system is trained to predict picking success via reinforcement learning, and in which training is done in a simulated picking environment.

166. The robotic fruit picking system of any preceding Claim in which the system is trained to predict picking success via reinforcement learning, in which a predictor of picking success is to ensure that the predicted path of the end effector sweeps through a 3D volume that encompasses the target stalk but not other stalks.

167. The robotic fruit picking system of any preceding Claim in which the control subsystem is trained via reinforcement learning that includes actions carried out by human operators.

168. The robotic fruit picking system of any preceding Claim in which a machine learning approach is used to train a model to predict yield forecast.

169. The robotic fruit picking system of any preceding Claim in which the system

records a map coordinate system location along with an image of all detected fruit and the recorded data is used to train a model to estimate a crop yield forecast.

170. The robotic fruit picking system of any preceding Claim in which the system uses picking success data obtained by working robots to learn and refine the parameters of a dynamically updateable statistical model for estimating picking success probability.

171. The robotic fruit picking system of any preceding Claim in which the system is operable to cut the stem of a fruit, in which the fruit is picked by first severing and gripping its stem, and the body of the fruit is removed from its stem in a subsequent operation.

172. The robotic fruit picking system of any preceding Claim in which the system operable to sever the fruit from its stem using a jet of compressed air, without requiring the handling of the body of the fruit.

173. The robotic fruit picking system of any preceding Claim in which the robot includes a collar, shaped to facilitate forcing the body of the fruit off the stem.

174. The robotic fruit picking system of any preceding Claim in which the system is operable to sever the stem from its fruit by using the inertia of the body of the target fruit to separate the body of the fruit from its stem.

175. The robotic fruit picking system of any preceding Claim in which the system is operable to cut or sever the stem of a fruit by moving the fruit or its stalk using a reciprocating back-and-forth motion of the fruit in the direction approximately perpendicular to its stalk or an oscillatory rotary motion with an axis of rotation approximately parallel to the stalk.

176. The robotic fruit picking system of any preceding Claim in which a path planning algorithm is used to model obstacles as probabilistic models of scene space occupancy by different types of obstacle with different material properties.

177. The robotic fruit picking system of any preceding Claim in which the end effector is operable to cut the stem of a fruit, in which the system includes a deformable end effector that is designed to deform under compressive force.

178. The robotic fruit picking system of any preceding Claim in which the system is operable to cut the stem of a fruit without handling the body of the fruit.

179. The robotic fruit picking system of any preceding Claim in which the robot is operable at night with a computer vision system that operates at night, and picks fruit when they are cooler and hence firmer to minimize bruising.

180. The robotic fruit picking system of any preceding Claim in which the end effector is operable to cut a fruit stem cleanly and without tearing to increase fruit productivity.

181. The robotic fruit picking system of any preceding Claim in which the quality control subsystem predicts the flavour or quality of a fruit and places the fruit in a specific storage container according to the flavour or quality prediction.

182. The robotic fruit picking system of Claim 181 in which the prediction of the flavour or quality of a fruit depends on the analysis of a growth trajectory data measured over time for the fruit.

183. A method of optimizing fruit yield prediction by imaging each fruit using the robotic fruit picking system of any preceding Claim 1 - 182, to determine ripeness or suitability for picking.

184. A method of optimizing fruit yield mapping across a fruit farm or multiple fruit farms, by imaging each fruit using the robotic fruit picking system of any preceding Claim 1 - 182, to determine ripeness or suitability for picking.

185. A method of maximizing fruit shelf life by using the robotic fruit picking system of any preceding Claim 1 - 182.

186. A method of selectively storing or punnetising the fruit with the optimal flavor or quality by using the robotic fruit picking system of any preceding Claim 1 - 182.

187. Fruit when picked using the robotic fruit picking system of any preceding Claim 1 - 182.

188. The fruit of Claim 187, being strawberries, including strawberries grown on a table.

189. The fruit of Claim 187, being raspberries.

190. The fruit of Claim 187, being apples, or pears, or peaches, or grapes, plums, cherries, or olives.