Robots for Kiwifruit Harvesting and Pollination

Jamie Bell

REVIEW 3 major objections 6 minor 252 references

Robots for Kiwifruit Harvesting and Pollination

T0 review · 3 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A stem-pushing gripper reached 81 percent of kiwifruit in a dense canopy.

desk verdict Solid engineering thesis with a genuine stem-pushing mechanism and 30 km of autonomous navigation, but the headline 81%-vs-66% reachability claim rests on a manual study and should be read as an upper bound, not a hard measurement. read the letter →

arxiv 2507.15484 v1 pith:UDE22JFC submitted 2025-07-21 cs.RO

Jamie Bell This is my paper · ORCID

classification cs.RO

keywords kiwifruitharvestingstempushingdetachmentroboticpollinationpergolaorchard3Dlidarnavigationinstancesegmentationautonomousdrivingpedestriandetection

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Working toward robotic kiwifruit harvesters that can pick 80 percent of available fruit, this thesis argues that a central bottleneck is mechanical access: the detachment mechanism must physically reach fruit inside the dense pergola canopy. The key result is a stem-pushing end effector that, in a hand-guided pickability study over 213 fruit, could reach 81 percent of fruit, where the previous approach-from-below mechanism reached only 66 percent. The same project produced targeted flower pollination from a moving spray boom, over 30 km of autonomous driving in kiwifruit orchards using 3D lidar, camera-based row following that matched lidar performance, and a safety system using high-visibility vests for pedestrian detection. If the reachability result carries over to fully autonomous operation, the mechanical side of the 80 percent harvest goal is no longer the limiting factor.

What carries the argument

The load-bearing object is the compact stem-pushing mechanism: a gripper with two paddles that holds the kiwifruit still, plus a pivoting pusher that rotates from one paddle to the other and pushes the stem along an arc around the point where the stem meets the fruit. This turns the previously demonstrated 'rotate the fruit upward' detachment action into 'rotate the stem against a held fruit', which is the same relative rotation at the stem attachment point but with a much smaller swept volume beside the fruit. Supporting it is the second harvester pipeline: instance segmentation of individual unobstructed fruit in Time-of-Flight intensity data, a percentile-square method that aims the target point near the top of the fruit so the pusher contacts the stem rather than the skin, a singular-value-decomposition calibration between the sensor and the five-degree-of-freedom arm, and a waypoint planner that keeps the end effector low during extension, raises it in front of the fruit, and then moves it horizontally into the picking pose.

What would settle it

Run the fully autonomous second harvester over the same marked 4 square metre, 213-fruit area and count fruit actually detached and collected: if the autonomous harvest rate does not exceed the original mechanism's 66 percent reachability, or falls far short of the 81 percent manual pickability, then the reachability gain does not yet translate into a harvesting gain. A second check would repeat the pickability study while removing each fruit after classifying it, to see whether the 81 percent drops when cluster geometry changes.

Watch

Extended reading notes

Core claim

The central claim is that kiwifruit can be detached by holding the fruit stationary between two gripper paddles and pushing the stem sideways just above the fruit, so that the stem rotates about the stem-fruit attachment point and shears off cleanly. This stem-pushing action needs less clear space beside the fruit than the previous approach-from-below mechanisms, and it can approach fruit from the side, so it can reach fruit obstructed from below by branches, wires, or beams. In a marked 4 square metre area containing 213 fruit, the compact stem pusher was judged pickable for 81 percent of fruit, compared with 66 percent for the original harvester mechanism; the larger stem pusher with cameras reached 77 percent. Detachment tests with the high-torque direct-drive configuration succeeded in 20 of 20 attempts, and the integrated harvester, using instance segmentation on Time-of-Flight data and a path planner that lifts the end effector in front of the fruit before moving in, picked fruit in the orchard, including fruit obstructed from below. The thesis also reports that the automated system's overall success was lower than the manual pickability figure, which it attributes to the remaining perception, calibration, and control errors.

Load-bearing premise

The reachability result depends on the assumption that manually moving the mechanism into picking poses by hand predicts what the automated vision, calibration, and control pipeline can achieve; the integrated harvester tests show that this assumption is not yet satisfied.

Editorial extensions

If this is right

If the 81 percent manual reachability transfers to automation, the mechanical-access bottleneck is removed and the remaining path to the 80 percent harvest goal is perception and control accuracy rather than end-effector geometry.
Side approach with a lift-in-front-of-the-fruit path can recover much of the roughly 23 percent of fruit the original harvester left behind because wires, branches, or beams blocked them from below.
High-torque stem pushing detaches fruit cleanly in 20 of 20 trials, so the detachment action itself can be made reliable before speed is improved with a more powerful motor.
Camera-only row following matched 3D lidar row following in tests, suggesting pergola-orchard navigation does not require the more expensive lidar sensor.
The high-visibility-vest detector uses the intensity channel of the same lidar used for navigation, so pedestrian safety monitoring can be added without extra sensing hardware.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The 81 percent figure is an upper bound: it was measured by a person moving the mechanism by hand, and the integrated harvester's lower success implies the autonomous vision and control pipeline, not the mechanism, currently sets the realized harvest rate.
The pickability study classified each fruit without removing previously classified fruit; in real clusters, removing one fruit shifts its neighbours, so dynamic field harvest rates may differ from the static 81 percent.
Stem pushing creates a new failure case, stems blocked by adjacent branches, that the original approach did not face; a hybrid that grips the fruit and pulls down using the branch as the pusher could recover some of these fruit.
The matched performance of camera and lidar row following suggests a lower-cost vision-only navigation configuration could be tested, with lidar reserved for safety functions such as pedestrian detection.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Worth a read if you work on orchard robotics, but don't take the 81% vs 66% number at face value. The abstract's headline comparison comes from a manual pickability study: someone moved the detachment mechanisms by hand and judged whether each of 213 fruit could be picked. The baseline for the previous system was a mock-up, with no validation against the actual mechanism, and the study deliberately ignored base pose, so the 81% figure is an idealized upper bound, not an autonomous harvest rate. The integrated harvester tests later in the thesis picked well under 80% of fruit. The stress-test note is right about these threats.

What the thesis does well: the stem-pushing detachment mechanism is genuinely new, and the detachment experiments with different gear ratios are clean and tell you something physical. The 3D lidar feature extraction methods for pergola orchards, tested against hand-labelled ground truth, are a solid contribution, and 30+ km of autonomous driving is a real field result. The thesis is unusually honest about failure modes, listing the canopy moving, adjacent fruit getting knocked off, and the system missing its project goal, which makes the engineering claims more credible. The ToF sensor calibration at 2.4 mm mean error is also careful work.

Soft spots beyond the pickability study: many CNN results use very small datasets (six training images for Mask R-CNN on ToF data), there are no error bars on the field trials, and the computer vision row following comparison against lidar involves only a few test runs. No code or data is released, which limits reproducibility.

All that said, this is a mixed thesis with an overstated headline and a large amount of solid, incremental engineering. The reachability comparison is suggestive but not established. The navigation work and the mechanism design stand on their own, and I would cite those parts. If the authors submitted a condensed version that fixes the pickability methodology and tempers the abstract, I would send it to peer review. The core mechanism and the navigation results deserve referee time.

Referee Report

3 major / 6 minor

Summary. The manuscript, based on a 2019 PhD thesis, reports on the development and field testing of robots for kiwifruit harvesting, artificial pollination, orchard navigation, and safety-related pedestrian detection. The harvesting work introduces several kiwifruit detachment mechanisms, the most developed being a stem-pushing end effector mounted on a Kuka YouBot arm. The central claim is that this mechanism could reach over 80% of kiwifruit in a cluttered canopy, versus below 70% for the prior mechanism, based on a manual pickability study of 213 fruit. The pollination work uses flower detection with spray booms and a dry pollination end effector; the navigation work develops 3D lidar and camera methods for row following, including over 30 km of autonomous driving; and the safety work includes a high-visibility-vest lidar detector with reported 100% true positives within 4 m. The thesis explicitly acknowledges that the project's overall 80% harvest and 90% flower-hit goals were not met.

Significance. If the central reachability claim is reliable, the stem-pushing mechanism would address a known bottleneck in robotic kiwifruit harvesting: access to fruit that are obstructed from below. The manuscript is rich in field data and engineering detail: a 213-fruit pickability study, detachment trials across three gear configurations, multiple orchard trials with the integrated harvester, 30+ km of autonomous driving, and a pedestrian detector with a strong reported result in a limited range. The thesis also provides honest failure analysis, including explicit categories of harvester failure and a statement that the 80% harvest goal was not met. These strengths make the work a useful contribution to agricultural robotics practice, provided the headline quantitative comparisons are properly qualified.

major comments (3)

[2.7.1, Table 14] The headline comparison (81% vs 66% pickability) rests on a study in which the original mechanism was represented by an undescribed 'mock-up', pickability was judged by hand placement by the experimenter, and the base pose of the harvesting system was explicitly disregarded. Because this comparison is the paper's central quantitative claim, the absence of (i) a description and validation of the mock-up against the actual mechanism, (ii) any inter-rater reliability or blinding protocol, and (iii) any account of how base pose affects the result means the 81% vs 66% difference is not established as a measurement of the mechanisms' relative capability. The manuscript should either report these controls, or reframe the claim as an idealised mechanism-level reachability bound and soften the abstract accordingly.
[2.7.3, Tables 16-17] The integrated harvester tests are the operational test of the reachability claim, yet the thesis states that the 80% harvest goal was not met and the integrated tests reported lower success rates than the pickability study. The gap between the 81% pickability figure and the achieved automated harvest rate is never quantified, so a reader cannot tell whether the shortfall comes from detection, calibration, path planning, arm control, or the detachment mechanism itself. The discussion should explicitly report the integrated success rate alongside the pickability figure and identify which pipeline stages account for the drop.
[2.6.2, Table 13] The Mask R-CNN fruit detection results in Table 13 are based on 10 test images (60 labelled images total), with no cross-validation because no hyperparameter tuning was performed; AP=0.88 at IoU 0.5 with zero detections in the small-object cell is a weak basis for the claim that unobstructed fruit are reliably detected. At minimum, report per-image confidence intervals, a larger test set, or explicit treatment of the small-object failure mode and its impact on the integrated harvest rate.

minor comments (6)

[Abstract] The abstract states 'over 80 percent' and 'less than 70 percent' as if they were established measurements; since they come from a single manual study reported in Table 14, the abstract should say 'in a manual pickability study' and report a confidence interval or at least the sample size.
[2.7.1] The criterion 'the action of picking without completing detachment' conflates reachability (can the mechanism be placed) with pickability (can detachment plausibly be completed); consider reporting separately the number of fruit that admitted a feasible pose and the number that also allowed the detachment action.
[2.7.1] The 'mock-up of the original kiwifruit harvesting mechanism' is not described or shown; a figure with dimensions and degrees of freedom, or a reference to a validation test against the real mechanism, would make the baseline comparison interpretable.
[2.3.3, Tables 3-4] The instance segmentation pilot used only 6 and 16 training images; the text should clearly label these as pilot demonstrations and avoid general statements about Mask R-CNN accuracy without a held-out set of realistic size.
[4.4.8] The claim of 'over 30 km of autonomous driving' is reported in a single sentence; provide a breakdown by orchard, platform, and test conditions, along with any failure events, so readers can assess the coverage and reliability of the result.
[4.7.9, Table 35] The statement that the computer vision row following 'worked as well as the 3D lidar row following method' should be accompanied by a statistical comparison (e.g., RMS error with confidence intervals and number of runs) rather than a visual comparison in Figure 175.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the reachability comparison is a direct empirical field measurement, not a derivation from fitted inputs or a self-citation chain.

full rationale

The headline claim—that the compact stem-pushing mechanism reached over 80% of fruit while the original mechanism reached under 70%—comes from the pickability study in Subsection 2.7.1 (Table 14). The paper explicitly describes the procedure as manually moving each mechanism into a picking pose and making a per-fruit pickable/not-pickable judgement, independent of the vision and control systems. That is a measurement protocol, not a derivation. The 81% and 66% figures are observed counts (172/213 and 140/213), not outputs of equations fitted to the same data, so the central comparison does not reduce by construction to its own inputs. The 'previous state of the art mechanism' is the authors' own earlier harvester, and one sentence notes its 66% pickability is similar to an earlier team result, but the comparison in the abstract is supported by the directly measured mock-up trial in the same study rather than by an unverified self-citation. The manual, subjective nature of the pickability judgements and the use of an unvalidated mock-up are legitimate concerns about measurement validity and experimental rigor, but they are not circularity: nothing in the paper defines 'pickable' in terms of the 80% goal, and no fitted parameter is renamed as a prediction. The harvesting, pollination, and navigation results are likewise empirical field tests with hand-labelled ground truth or direct success/failure counts. There is no load-bearing self-citation, no uniqueness theorem imported from the authors' prior work, and no ansatz smuggled in via citation. The claimed derivation chain is therefore self-contained; the central figures stand or fall on the quality of the field measurements, not on circular reasoning.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claims rest on measured field performance and standard ML training; the main free parameters are hand-chosen thresholds for lidar feature extraction and boom navigation. No new physical entities are postulated. The most significant assumption is that manual pickability predicts automated harvesting success.

free parameters (4)

Nearest neighbour separation threshold for row detection
Used in the 3D lidar row detection algorithm (Subsection 4.4.1) to pair post/trunk clusters; value chosen by hand, affects row centreline accuracy.
Number of angular segments for plane selection
In 3D lidar feature extraction (Subsection 4.5.4), the number of segments and the plane selection metric are chosen by hand; precision/recall varies with segment count.
Object height threshold for vertical connected objects
In the vertical connected objects feature extraction method (Subsection 4.5.6), height thresholds separate structure from canopy; results vary with threshold.
Hyperparameters for flower detection FCN-8s = selected by trial and error
The thesis states in Table 21 that hyperparameters were selected by trial and error; they affect the dry pollination flower detection performance.

assumptions (4)

domain assumption An unobstructed fruit in the sensor image is also unobstructed for the robot arm.
Subsection 2.6.2 explicitly states this assumption for the second kiwifruit harvester; it underpins the strategy of only targeting unobstructed fruit.
domain assumption The manual pickability assessment is representative of automated performance.
Subsection 2.7.1 uses hand placement to judge reachability; this is load-bearing for the 81% reachability claim.
domain assumption The pergola structure is sufficiently consistent for 3D lidar feature extraction methods to generalize across orchards.
The navigation feature extraction methods (Section 4.5) are tested in specific orchards; generalization is assumed for broader deployment.
domain assumption The performance of CNNs trained on small datasets is representative enough to judge the method.
Several detection models are trained on very small datasets (e.g., 6 training images for Mask R-CNN in Subsection 2.3.3); the thesis argues transfer learning and pixel-level labels compensate, but this is an assumption.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Robots for Kiwifruit Harvesting and Pollination." pith.science (2026). https://pith.science/paper/UDE22JFC

@misc{pith2026250715484,
  author       = {Pith},
  title        = {Pith review of: Robots for Kiwifruit Harvesting and Pollination},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UDE22JFC}},
  note         = {Machine review of arXiv:2507.15484}
}

read the original abstract

This research was a part of a project that developed mobile robots that performed targeted pollen spraying and automated harvesting in pergola structured kiwifruit orchards. Multiple kiwifruit detachment mechanisms were designed and field testing of one of the concepts showed that the mechanism could reliably pick kiwifruit. Furthermore, this kiwifruit detachment mechanism was able to reach over 80 percent of fruit in the cluttered kiwifruit canopy, whereas the previous state of the art mechanism was only able to reach less than 70 percent of the fruit. Artificial pollination was performed by detecting flowers and then spraying pollen in solution onto the detected flowers from a line of sprayers on a boom, while driving at up to 1.4 ms-1. In addition, the height of the canopy was measured and the spray boom was moved up and down to keep the boom close enough to the flowers for the spray to reach the flowers, while minimising collisions with the canopy. Mobile robot navigation was performed using a 2D lidar in apple orchards and vineyards. Lidar navigation in kiwifruit orchards was more challenging because the pergola structure only provides a small amount of data for the direction of rows, compared to the amount of data from the overhead canopy, the undulating ground and other objects in the orchards. Multiple methods are presented here for extracting structure defining features from 3D lidar data in kiwifruit orchards. In addition, a 3D lidar navigation system -- which performed row following, row end detection and row end turns -- was tested for over 30 km of autonomous driving in kiwifruit orchards. Computer vision algorithms for row detection and row following were also tested. The computer vision algorithm worked as well as the 3D lidar row following method in testing.

Figures

Figures reproduced from arXiv: 2507.15484 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

252 extracted references · 76 canonical work pages

[1]

Then the overlaps between calyx instances and skin instances were found

In order to find calyx-skin pairs, the calyx segmentation outputs were dilated. Then the overlaps between calyx instances and skin instances were found. For each calyx instance, the skin instance with the maximum overlap was assigned to the calyx instance
[2]

The centre of each calyx instance was found using the median x-coordinate and y-coordinate in the calyx segmentation output. 273
[3]

For each calyx-skin pair, Principal Components Analysis (PCA) of the skin instance points was used to find the centre of the skin and the directions of the principal components
[4]

It was thought that this procedure might work well for unoccluded fruit

For each calyx-skin pair, the calyx centre was reflected about the second principal component. It was thought that this procedure might work well for unoccluded fruit. However, for occluded fruit it was postulated that PCA might not find the principal components of the fruit accurately. Although, for occluded fruit, it was thought that the collision detec...
[5]

turn left

Repeating steps 1-4 multiple times. The camera images collected were post-processed into 2 classes for classification; “turn left” and “turn right”. The resulting data was used to train an AlexNet CNN, using the hyperparameters given in Table 53. The resulting average validation accuracy from 3 training runs was 88%. This result 277 seemed quite poor for ...

2016
[6]

A fixed size region of interest around the centre of a detected fruit was set
[7]

The points in the region of interest were transformed to the robot arm coordinate system
[8]

The points were transformed into the coordinate system of the gripper, for the pose of the gripper at the instant of fruit detachment

Show all 252 references

[9]

soft collision

Points within the fixed volume of the gripper were classified as in collision. It was decided that the output for collision checking should be a single score for each fruit instance. This score was supposed to represent the current risk from collisions, if the fruit was picked...
[10]

Manually driving to create a linear and angular offset to correct
[11]

Each image was saved with the steering command that was being given at the same time, as well as a sequence identification number to ensure that the name of each image was unique

Pushing a button on the joystick to trigger the recording of images. Each image was saved with the steering command that was being given at the same time, as well as a sequence identification number to ensure that the name of each image was unique
[12]

Manually driving to correct the linear and angular offset
[13]

Pushing another button on the joystick to stop the recording of images
[14]

Autonomous Fruit Picking Machine: A Robotic Apple Harvester,

J. Baeten, K. Donné, S. Boedrij, W. Beckers, and E. Claesen, “Autonomous Fruit Picking Machine: A Robotic Apple Harvester,” in Field and Service Robotics: Results of the 6th International Conference, C. Laugier and R. Siegwart, Eds. Berlin, Heidelberg: Springer Berlin Heidelbe...

2008
[15]

International Organization for Standardization, 2015

ISO/TC 199, ISO 13849-1:2015, Safety of machinery -- Safety-related parts of control systems -- Part 1: General principles for design. International Organization for Standardization, 2015

2015
[16]

The PCD (Point Cloud Data) file format

“The PCD (Point Cloud Data) file format.” [Online]. Available: http://pointclouds.org/documentation/tutorials/pcd_file_format.php. [Accessed: 01-Jun- 2019]

2019
[17]

International Electrotechnical Commission, 2010

IEC, IEC 61508 Functional safety of electrical/electronic/programmable electronic safety- related systems. International Electrotechnical Commission, 2010

2010
[18]

You Only Look Once: Unified, Real-Time Object Detection,

J. Redmon, S. K. Divvala, R. B. Girshick, and A. Farhadi, “You Only Look Once: Unified, Real-Time Object Detection,” CoRR, vol. abs/1506.0, 2015

2015
[19]

A. G. Aitken and E. W. Hewett, Fresh Facts. Auckland: The New Zealand Institute for Plant & Food Research Ltd, 2015

2015
[20]

Timmins, Seasonal Employment Patterns in the Horticultural Industry

J. Timmins, Seasonal Employment Patterns in the Horticultural Industry. Wellington: Statistics New Zealand, 2009

2009
[21]

Rural Delivery, Series 11, Episode 38,

K. Cooper and J. Cvitanovich, “Rural Delivery, Series 11, Episode 38,” 2015

2015
[22]

Hyundai Country Calendar, Series 2015, Episode 19,

D. Henry, “Hyundai Country Calendar, Series 2015, Episode 19,” 2015

2015
[23]

Rural Delivery, Series 11, Episode 34,

K. Cooper and J. Cvitanovich, “Rural Delivery, Series 11, Episode 34,” 2015

2015
[24]

Hyundai Country Calendar, Series 2014, Episode 19,

D. Henry, “Hyundai Country Calendar, Series 2014, Episode 19,” 2014

2014
[25]

Hyundai Country Calendar, Series 2015, Episode 17,

D. Henry, “Hyundai Country Calendar, Series 2015, Episode 17,” 2015

2015
[26]

Robotic Apple Harvesting in Washington State

J. Davidson and A. Silwal, “Robotic Apple Harvesting in Washington State.” [Online]. Available: ws.cubbyusercontent.com/p/_325a12760eb94449b4faa172f1db5e56/AgRA+Webinar+_Robo tic+Apple+Harvesting+in+Washington+State_+2015-12-15.webm/1563073681. [Accessed: 05-Apr-2016]

2015
[27]

Computer Vision for Fruit Harvesting Robots; State of the Art and Challenges Ahead,

K. Kapach, E. Barnea, R. Mairon, Y . Edan, and O. Ben-Shahar, “Computer Vision for Fruit Harvesting Robots; State of the Art and Challenges Ahead,” Int. J. Comput. Vis. Robot., vol. 3, no. 1/2, pp. 4–34, Apr. 2012

2012
[28]

Automated Crop Yield Estimation for Apple Orchards,

Q. Wang, S. Nuske, M. Bergerman, and S. Singh, “Automated Crop Yield Estimation for Apple Orchards,” Proc. Int. Symp. Exp. Robot., 2012

2012
[29]

Goodwin, Pollination of Crops in Australia and New Zealand

M. Goodwin, Pollination of Crops in Australia and New Zealand. Canberra: Rural Industries Research and Development Corporation, 2012. 281

2012
[30]

Mount Maunganui: New Zealand Kiwifruit Growers Incorporated, 2016

NZKGI, 2016 Kiwifruit Book: A resource for New Zealand secondary school teachers and growers new to the kiwifruit industry. Mount Maunganui: New Zealand Kiwifruit Growers Incorporated, 2016

2016
[31]

M. J. W. Hii, Kiwifruit Flower Pollination: Wind Pollination Efficiencies and Sprayer Jet Applications, Doctor of Philosophy, Chemical and Process Engineering. Canterbury: University of Canterbury, 2004

2004
[32]

A review of automation and robotics for the bioindustry,

T. Grift, Q. Zhang, N. Kondo, and K. C. Ting, “A review of automation and robotics for the bioindustry,” J. Biomechatronics Eng., vol. 1, no. 1, pp. 37–54, 2008

2008
[33]

Garnett, What is a Sustainable Healthy Diet? A Discussion Paper

T. Garnett, What is a Sustainable Healthy Diet? A Discussion Paper. London: Food Climate Research Network, 2014

2014
[34]

Ministry declares 1200-person kiwifruit labour shortage in Bay of Plenty,

S. Motion, “Ministry declares 1200-person kiwifruit labour shortage in Bay of Plenty,” NZ Herald, 08-May-2018

2018
[35]

A. J. Scarfe, Development of an autonomous kiwifruit harvester : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Industrial Automation at Massey University, Manawatu, New Zealand. Manawatu: Massey University, 2012

2012
[36]

Kiwifruit recognition at nighttime using artificial lighting based on machine vision,

F. Longsheng, W. Bin, C. Yongjie, S. Shuai, Y . Gejima, and T. Kobayashi, “Kiwifruit recognition at nighttime using artificial lighting based on machine vision,” Int. J. Agric. Biol. Eng., vol. 8, no. 4, p. 52, 2015

2015
[37]

A Method for Separation of Kiwifruit Adjacent Fruits Based on Hough Transformation,

Y . Cui, S. Su, Z. Lv, P. Li, and X. Ding, “A Method for Separation of Kiwifruit Adjacent Fruits Based on Hough Transformation,” J. Agric. Mech. Res., pp. 166–169, 2012

2012
[38]

Research on the Object Extraction of Kiwifruit Based on Machine Vision,

T. Wu, C. Yuan, and J. Chen, “Research on the Object Extraction of Kiwifruit Based on Machine Vision,” J. Agric. Mech. Res., pp. 21–26, 2012

2012
[39]

Design and control of an apple harvesting robot,

Z. De-An, L. Jidong, J. Wei, Z. Ying, and C. Yu, “Design and control of an apple harvesting robot,” Biosyst. Eng., vol. 110, no. 2, pp. 112–122, 2011

2011
[40]

Apple crop-load estimation with over-the-row machine vision system,

A. Gongal, A. Silwal, S. Amatya, M. Karkee, Q. Zhang, and K. Lewis, “Apple crop-load estimation with over-the-row machine vision system,” Comput. Electron. Agric., vol. 120, pp. 26–35, 2016

2016
[41]

Apple detection algorithm for robotic harvesting using a RGB-D camera,

T. T. Nguyen, K. Vandevoorde, E. Kayacan, J. De Baerdemaeker, and W. Saeys, “Apple detection algorithm for robotic harvesting using a RGB-D camera,” in Proceedings of the International Conference of Agricultural Engineering, 2014

2014
[42]

Rich feature hierarchies for accurate object detection and semantic segmentation,

R. B. Girshick, J. Donahue, T. Darrell, and J. Malik, “Rich feature hierarchies for accurate object detection and semantic segmentation,” CoRR, vol. abs/1311.2, 2013

2013
[43]

Low and high-level visual feature- based apple detection from multi-modal images,

J. P. Wachs, H. I. Stern, T. Burks, and V . Alchanatis, “Low and high-level visual feature- based apple detection from multi-modal images,” Precis. Agric., vol. 11, no. 6, 2010. 282

2010
[44]

Kiwifruit yield estimation using image processing by an Android mobile phone,

L. Fu, Z. Liu, Y . Majeed, and Y . Cui, “Kiwifruit yield estimation using image processing by an Android mobile phone,” IF AC-PapersOnLine, vol. 51, no. 17, pp. 185–190, 2018

2018
[45]

A novel image processing algorithm to separate linearly clustered kiwifruits,

L. Fu, E. Tola, A. Al-Mallahi, R. Li, and Y . Cui, “A novel image processing algorithm to separate linearly clustered kiwifruits,” Biosyst. Eng., vol. 183, pp. 184–195, 2019

2019
[46]

ImageNet Classification with Deep Convolutional Neural Networks,

A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet Classification with Deep Convolutional Neural Networks,” in Advances in Neural Information Processing Systems 25, F. Pereira, C. J. C. Burges, L. Bottou, and K. Q. Weinberger, Eds. Curran Associates, Inc., 2012, pp. 1097–1105

2012
[47]

Learning Hierarchical Features for Scene Labeling,

C. Farabet, C. Couprie, L. Najman, and Y . LeCun, “Learning Hierarchical Features for Scene Labeling,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 35, no. 8, pp. 1915–1929, Aug. 2013

1915
[48]

Going Deeper with Convolutions,

C. Szegedy, W. Liu, Y . Jia, P. Sermanet, S. E. Reed, D. Anguelov, D. Erhan, V . Vanhoucke, and A. Rabinovich, “Going Deeper with Convolutions,” CoRR, vol. abs/1409.4, 2014

2014
[49]

Deep learning,

Y . LeCun, Y . Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436– 444, May 2015

2015
[50]

Kiwifruit detection in field images using Faster R-CNN with ZFNet,

L. Fu, Y . Feng, Y . Majeed, X. Zhang, J. Zhang, M. Karkee, and Q. Zhang, “Kiwifruit detection in field images using Faster R-CNN with ZFNet,” IF AC-PapersOnLine, vol. 51, no. 17, pp. 45–50, 2018

2018
[51]

Deep Fruit Detection in Orchards,

S. Bargoti and J. P. Underwood, “Deep Fruit Detection in Orchards,” CoRR, vol. abs/1610.0, 2016

2016
[52]

Image Based Mango Fruit Detection, Localisation and Yield Estimation Using Multiple View Geometry,

M. Stein, S. Bargoti, and J. Underwood, “Image Based Mango Fruit Detection, Localisation and Yield Estimation Using Multiple View Geometry,” Sensors (Basel), vol. 16, no. 11, 2016

2016
[53]

DeepFruits: A Fruit Detection System Using Deep Neural Networks,

I. Sa, Z. Ge, F. Dayoub, B. Upcroft, T. Perez, and C. McCool, “DeepFruits: A Fruit Detection System Using Deep Neural Networks,” Sensors, vol. 16, no. 8, p. 1222, 2016

2016
[55]

Fast {R-CNN},

R. B. Girshick, “Fast {R-CNN},” CoRR, vol. abs/1504.0, 2015

2015
[56]

Design of End-effector for Kiwifruit Harvesting Robot Experiment,

L. Mu, Y . Liu, Y . Cui, H. Liu, L. Chen, L. Fu, and G. Yoshinori, “Design of End-effector for Kiwifruit Harvesting Robot Experiment,” 2017 ASABE Annual International Meeting. ASABE, St. Joseph, MI, p. 1, 2017

2017
[57]

You Only Look Once: Unified, Real-Time Object Detection,

“You Only Look Once: Unified, Real-Time Object Detection,” ComputerVisionFoundation Videos, 2016. [Online]. Available: www.youtube.com/watch?v=NM6lrxy0bxs. [Accessed: 21-Jun-2017]

2016
[58]

DetectNet: Deep Neural Network for Object Detection in DIGITS,

A. Tao, J. Barker, and S. Sarathy, “DetectNet: Deep Neural Network for Object Detection in DIGITS,” 2016. [Online]. Available: devblogs.nvidia.com/parallelforall/detectnet-deep- neural-network-object-detection-digits/. [Accessed: 24-May-2017]. 283

2016
[59]

Fully Convolutional Networks for Semantic Segmentation,

E. Shelhamer, J. Long, and T. Darrell, “Fully Convolutional Networks for Semantic Segmentation,” CoRR, vol. abs/1605.0, 2016

2016
[60]

Gradient-based learning applied to document recognition,

Y . LeCun, L. Bottou, Y . Bengio, and P. Haffner, “Gradient-based learning applied to document recognition,” Proc. IEEE, vol. 86, no. 11, pp. 2278–2324, 1998

1998
[61]

A MultiPath Network for Object Detection,

S. Zagoruyko, A. Lerer, T.-Y . Lin, P. H. O. Pinheiro, S. Gross, S. Chintala, and P. Dollár, “A MultiPath Network for Object Detection,” CoRR, vol. abs/1604.0, 2016

2016
[62]

Mask {R-CNN},

K. He, G. Gkioxari, P. Dollár, and R. B. Girshick, “Mask {R-CNN},” CoRR, vol. abs/1703.0, 2017

2017
[63]

Simple Stereo Matching Algorithm for Localising Keypoints in a Restricted Search Space,

M. Seabright, L. Streeter, M. Cree, M. Duke, and R. Tighe, “Simple Stereo Matching Algorithm for Localising Keypoints in a Restricted Search Space,” in 2018 International Conference on Image and Vision Computing New Zealand (IVCNZ), 2018, pp. 1–6

2018
[64]

A Visual System of Citrus Picking Robot Using Convolutional Neural Networks,

Y . Liu, C. Yang, H. Ling, S. Mabu, and T. Kuremoto, “A Visual System of Citrus Picking Robot Using Convolutional Neural Networks,” in 2018 5th International Conference on Systems and Informatics (ICSAI), 2018, pp. 344–349

2018
[65]

Convolutional Pose Machines,

S.-E. Wei, V . Ramakrishna, T. Kanade, and Y . Sheikh, “Convolutional Pose Machines,” CoRR, vol. abs/1602.00134, 2016

2016 arXiv
[66]

Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields,

Z. Cao, T. Simon, S.-E. Wei, and Y . Sheikh, “Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields,” in CVPR, 2017

2017
[67]

Development and experiment of end-effector for kiwifruit harvesting robot,

L. Fu, F. Zhang, Y . Gejima, Z. Li, B. Wang, and Y . Cui, “Development and experiment of end-effector for kiwifruit harvesting robot,” Trans Chin Soc Agric Mach, vol. 46, no. 3, pp. 1–8, 2015

2015
[68]

Design of end-effector for kiwifruit harvesting robot,

J. Chen, H. Wang, H. Jiang, H. Gao, W. Lei, and G. Dang, “Design of end-effector for kiwifruit harvesting robot,” Nongye Jixie Xuebao/Transactions Chinese Soc. Agric. Mach., vol. 43, no. 10, pp. 151–154, 2012

2012
[69]

Development and experiment of end-effector for kiwifruit harvesting robot,

L. Fu, F. Zhang, G. Yoshinori, Z. Li, B. Wang, and Y . Cui, “Development and experiment of end-effector for kiwifruit harvesting robot,” Nongye Jixie Xuebao/Transactions Chinese Soc. Agric. Mach., vol. 46, no. 3, pp. 1–8, 2015

2015
[70]

Colour-agnostic shape-based 3D fruit detection for crop harvesting robots,

E. Barnea, R. Mairon, and O. Ben-Shahar, “Colour-agnostic shape-based 3D fruit detection for crop harvesting robots,” Biosyst. Eng., vol. 146, pp. 57–70, 2016

2016
[71]

Design and simulation of an integrated end-effector for picking kiwifruit by robot,

L. Mu, G. Cui, Y . Liu, Y . Cui, L. Fu, and Y . Gejima, “Design and simulation of an integrated end-effector for picking kiwifruit by robot,” Inf. Process. Agric., 2019

2019
[72]

Robotic kiwifruit harvesting using machine vision, convolutional neural networks, and robotic arms,

H. A. M. Williams, M. H. Jones, M. Nejati, M. J. Seabright, J. Bell, N. D. Penhall, J. J. Barnett, M. D. Duke, A. J. Scarfe, H. S. Ahn, J. Lim, and B. A. MacDonald, “Robotic kiwifruit harvesting using machine vision, convolutional neural networks, and robotic arms,” Biosyst. E...

2019
[73]

Improvements to and large-scale evaluation of a robotic kiwifruit harvester,

H. Williams, C. Ting, M. Nejati, M. H. Jones, N. Penhall, J. Lim, M. Seabright, J. Bell, H. S. Ahn, A. Scarfe, M. Duke, and B. MacDonald, “Improvements to and large-scale evaluation of a robotic kiwifruit harvester,” J. F . Robot., vol. 0, no. 0, 2019

2019
[74]

Barnett, Prismatic axis, differential-drive robotic kiwifruit harvester for reduced cycle time

J. Barnett, Prismatic axis, differential-drive robotic kiwifruit harvester for reduced cycle time. A thesis submitted in fulfillment of the requirements for the degree of Master of Engineering at The University of Waikato. Hamilton: The University of Waikato, 2018

2018
[75]

Design and Testing of a Kiwifruit Harvester End-Effector,

S. S. Graham, W. Zong, J. Feng, and S. Tang, “Design and Testing of a Kiwifruit Harvester End-Effector,” Trans. ASABE, vol. 61, no. 1, pp. 45–51, 2018

2018
[76]

C. Ting, J. Barnett, and M. Duke, Kiwifruit detachment methods. Hamilton: The University of Waikato, 2018

2018
[77]

A cherry-tomato harvesting robot,

F. Taqi, F. Al-Langawi, H. Abdulraheem, and M. El-Abd, “A cherry-tomato harvesting robot,” in 2017 18th International Conference on Advanced Robotics (ICAR), 2017, pp. 463– 468

2017
[78]

End effector for robotic harvesting,

C. Salisbury, J. Suchoski, and R. Mahoney, “End effector for robotic harvesting,” US20170273241A1, 2014

2014
[79]

Evaluation of a localized shake-and-catch harvesting system for fresh market apples,

L. He, H. Fu, H. Xia, M. Karkee, Q. Zhang, and M. Whiting, “Evaluation of a localized shake-and-catch harvesting system for fresh market apples,” Agric. Eng. Int. CIGR J., vol. 19, no. 4, pp. 36–44, 2017

2017
[80]

Development of an autonomous kiwifruit picking robot,

A. J. Scarfe, R. C. Flemmer, H. H. C. Bakker, and C. L. Flemmer, “Development of an autonomous kiwifruit picking robot,” in 4th International Conference on Autonomous Robots and Agents, {ICARA} 2009, Wellington, New Zealand, February 10-12, 2009, 2009, pp. 380– 384

2009
[81]

Sensors and systems for fruit detection and localization: A review,

A. Gongal, S. Amatya, M. Karkee, Q. Zhang, and K. Lewis, “Sensors and systems for fruit detection and localization: A review,” Comput. Electron. Agric., vol. 116, no. July, pp. 8–19, 2015

2015
[82]

Review on fruit harvesting method for potential use of automatic fruit harvesting systems,

P. Li, S. Lee, and H.-Y . Hsu, “Review on fruit harvesting method for potential use of automatic fruit harvesting systems,” Procedia Eng., vol. 23, pp. 351–366, 2011

2011
[83]

Development of a tomato harvesting robot used in greenhouse,

L. Wang, B. Zhao, J. Fan, X. Hu, S. Wei, Y . Li, Q. Zhou, and C. Wei, “Development of a tomato harvesting robot used in greenhouse,” Int. J. Agric. Biol. Eng., vol. 10, no. 4, 2017

2017
[84]

Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors,

J. Huang, V . Rathod, C. Sun, M. Zhu, A. Korattikara, A. Fathi, I. Fischer, Z. Wojna, Y . Song, S. Guadarrama, and K. Murphy, “Speed/Accuracy Trade-Offs for Modern Convolutional Object Detectors,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017,...

2017
[85]

Automatic apple recognition based on the fusion of color and 3D feature for robotic fruit picking,

Y . Tao and J. Zhou, “Automatic apple recognition based on the fusion of color and 3D feature for robotic fruit picking,” Comput. Electron. Agric., vol. 142, pp. 388–396, 2017

2017
[86]

Detection of red and bicoloured apples on tree with an RGB-D camera,

T. T. Nguyen, K. Vandevoorde, N. Wouters, E. Kayacan, J. G. De Baerdemaeker, and W. Saeys, “Detection of red and bicoloured apples on tree with an RGB-D camera,” Biosyst. Eng., vol. 146, pp. 33–44, 2016. 285

2016
[87]

Development of a Tomato Harvesting Robot,

S. Yasukawa, B. Li, T. Sonoda, and K. Ishii, “Development of a Tomato Harvesting Robot,” in The 2017 International Conference on Artificial Life and Robotics (ICAROB 2017), 2017, pp. 408–411

2017
[88]

C920 HD Pro Webcam

“C920 HD Pro Webcam.” [Online]. Available: www.logitech.com/en-nz/product/hd-pro- webcam-c920#specification-tabular. [Accessed: 02-Jun-2019]

2019
[89]

acA1920-40uc - Basler ace

“acA1920-40uc - Basler ace.” [Online]. Available: www.baslerweb.com/en/products/cameras/area-scan-cameras/ace/aca1920-40uc/. [Accessed: 02-Jun-2019]

2019
[90]

Meet Zed

“Meet Zed.” [Online]. Available: www.stereolabs.com/zed/. [Accessed: 02-Jun-2019]

2019
[91]

tof640-20gm_850nm - Basler Time-of-Flight,

“tof640-20gm_850nm - Basler Time-of-Flight,” 2019. [Online]. Available: www.baslerweb.com/en/products/cameras/3d-cameras/time-of-flight-camera/tof640- 20gm_850nm/. [Accessed: 03-May-2019]

2019
[92]

Evaluation of 3D-Sensorsystems for service robotics in orcharding and viticulture,

E. Wunder, A. Linz, A. Trabhardt, and A. Ruckelshausen, “Evaluation of 3D-Sensorsystems for service robotics in orcharding and viticulture,” in 72nd International Conference of Agricultural Engineering, 2014, pp. 83–88

2014
[93]

Design, integration, and field evaluation of a robotic apple harvester,

A. Silwal, J. R. Davidson, M. Karkee, C. Mo, Q. Zhang, and K. Lewis, “Design, integration, and field evaluation of a robotic apple harvester,” J. F . Robot., vol. 34, no. 6, pp. 1140–1159, 2017

2017
[94]

Available: https://lenses.kowa-usa.com/hc-series/417-lm6hc.html

“LM6HC.” [Online]. Available: https://lenses.kowa-usa.com/hc-series/417-lm6hc.html. [Accessed: 02-Jun-2019]

2019
[95]

Caffe: Convolutional Architecture for Fast Feature Embedding,

Y . Jia, E. Shelhamer, J. Donahue, S. Karayev, J. Long, R. B. Girshick, S. Guadarrama, and T. Darrell, “Caffe: Convolutional Architecture for Fast Feature Embedding,” CoRR, vol. abs/1408.5, 2014

2014
[96]

Nvidia Digits

“Nvidia Digits.” [Online]. Available: https://developer.nvidia.com/digits. [Accessed: 08- May-2017]

2017
[97]

bvlc_alexnet

“bvlc_alexnet.” [Online]. Available: https://github.com/BVLC/caffe/tree/master/models/bvlc_alexnet. [Accessed: 18-Mar-2017]

2017
[98]

Wind and honey bee pollination of kiwifruit (Actinidia chinensis ‘HORT16A’),

R. M. Goodwin, H. M. McBrydie, and M. A. Taylor, “Wind and honey bee pollination of kiwifruit (Actinidia chinensis ‘HORT16A’),” New Zeal. J. Bot., vol. 51, no. 3, pp. 229–240, 2013

2013
[99]

Tensorflow Object Detection API

“Tensorflow Object Detection API.” [Online]. Available: https://github.com/tensorflow/models/tree/master/research/object_detection. [Accessed: 30- Apr-2019]

2019
[100]

COCO Common Objects in Context, Detection Evaluation

“COCO Common Objects in Context, Detection Evaluation.” [Online]. Available: cocodataset.org/#detection-eval. [Accessed: 21-Apr-2019]. 286

2019
[101]

Tensorflow detection model zoo

“Tensorflow detection model zoo.” [Online]. Available: https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection _model_zoo.md. [Accessed: 02-Jun-2019]

2019
[102]

YouBot Detailed Specifications,

“YouBot Detailed Specifications,” 2015. [Online]. Available: http://www.youbot- store.com/wiki/index.php/YouBot_Detailed_Specifications. [Accessed: 02-Jun-2019]

2015
[103]

Classification of Kiwifruit Grades Based on Fruit Shape Using a Single Camera,

L. Fu, S. Sun, R. Li, and S. Wang, “Classification of Kiwifruit Grades Based on Fruit Shape Using a Single Camera,” Sensors (Basel)., vol. 16, no. 7, p. 1012, Jul. 2016

2016
[104]

Ting, Comparison between Selective and Non-selective Kiwifruit Harvesters for Economic Viability

C. Ting, Comparison between Selective and Non-selective Kiwifruit Harvesters for Economic Viability. A thesis submitted in fulfilment of the requirements for the degree of Master of Engineering at The University of Waikato by Chia-nan Ting. Hamilton: The University of Waikato, 2018

2018
[105]

Model L-15 DC Solenoid Tubular, Pull Type

“Model L-15 DC Solenoid Tubular, Pull Type.” [Online]. Available: https://media.digikey.com/pdf/Data Sheets/Pontiac Coil Inc PDFs/L-15.pdf. [Accessed: 03- May-2019]

2019
[106]

daA1600-60uc (S-Mount) - Basler dart,

“daA1600-60uc (S-Mount) - Basler dart,” 2019. [Online]. Available: www.baslerweb.com/en/products/cameras/area-scan-cameras/dart/daa1600-60uc-s-mount/. [Accessed: 03-May-2019]

2019
[107]

Intel RealSense Depth Camera D400-Series

“Intel RealSense Depth Camera D400-Series.” [Online]. Available: https://software.intel.com/en-us/realsense/d400. [Accessed: 03-May-2019]

2019
[108]

Kaehler and G

A. Kaehler and G. Bradski, Learning OpenCV 3: Computer Vision in C++ with the OpenCV Library. Sebastopol: O’Reilly Media, 2016

2016
[109]

Motion Planning,

L. E. Kavraki and S. M. LaValle, “Motion Planning,” in Springer Handbook of Robotics, 2nd ed., B. Siciliano and O. Khatib, Eds. Switzerland: Springer International Publishing, 2016, pp. 139–162

2016
[110]

Husky Unmanned Ground Vehicle

“Husky Unmanned Ground Vehicle.” [Online]. Available: www.clearpathrobotics.com/husky-unmanned-ground-vehicle-robot/. [Accessed: 02-Jun- 2019]

2019
[111]

GEFORCE GTX 1080 Ti

“GEFORCE GTX 1080 Ti.” [Online]. Available: www.nvidia.com/en- us/geforce/products/10series/geforce-gtx-1080-ti/. [Accessed: 02-Jun-2019]

2019
[112]

Dynamics and flight control of a flapping-wing robotic insect in the presence of wind gusts,

P. Chirarattananon, Y . Chen, E. F. Helbling, K. Y . Ma, R. Cheng, and R. J. Wood, “Dynamics and flight control of a flapping-wing robotic insect in the presence of wind gusts,” Interface Focus, vol. 7, no. 1, 2016

2016
[113]

Pollination Effectiveness of Honey Bees (Hymenoptera: Apidae) in a Kiwifruit Orchard,

B. E. VaissiÉre, G. Rodet, M. Cousin, L. Botella, and J.-P. T. Grossa, “Pollination Effectiveness of Honey Bees (Hymenoptera: Apidae) in a Kiwifruit Orchard,” J. Econ. Entomol., vol. 89, no. 2, pp. 453–461, 1996. 287

1996
[114]

The effect of kiwifruit (Actinidia deliciosa A Chev) and yellow flowered broom (Cytisus scoparius Link) pollen on the ovary development of worker honey bees (Apis mellifera L),

Jay, S. C. and Jay, D. H., “The effect of kiwifruit (Actinidia deliciosa A Chev) and yellow flowered broom (Cytisus scoparius Link) pollen on the ovary development of worker honey bees (Apis mellifera L),” Apidologie, vol. 24, no. 6, pp. 557–563, 1993

1993
[115]

Kiwifruit pollination by honey bees, Tauranga observations, 1978–81,

P. G. Clinch, “Kiwifruit pollination by honey bees, Tauranga observations, 1978–81,” New Zeal. J. Exp. Agric., vol. 12, no. 1, pp. 29–38, 1984

1978
[116]

R. M. Goodwin, A. T. Houten, and J. Perry, MAF Report on the Evaluation of the “POLLI” Pollination Device. Wellington: Ministry of Agriculture and Forestry, New Zealand, 1991

1991
[117]

Artificial Pollination in Kiwifruit and Olive Trees,

T. Gianni, “Artificial Pollination in Kiwifruit and Olive Trees,” M. V . E.-P. W. Mokwala, Ed. Rijeka: IntechOpen, 2018, p. Ch. 5

2018
[118]

Hand and Machine Pollination of Kiwifruit,

B. Razeto, G. Reginato, and A. Larraín, “Hand and Machine Pollination of Kiwifruit,” Int. J. Fruit Sci., vol. 5, no. 2, pp. 37–44, 2005

2005
[119]

QuadDuster,

“QuadDuster,” 2019. [Online]. Available: www.roboticsplus.co.nz/quadduster. [Accessed: 19-Mar-2019]

2019
[120]

Pollensmart,

“Pollensmart,” 2018. [Online]. Available: www.pollensmart.co.nz/. [Accessed: 19-Mar- 2019]

2018
[121]

Precision pollination,

M. Hansen, “Precision pollination,” Good Fruit Grower, 2015. [Online]. Available: www.goodfruit.com/precision-pollination/. [Accessed: 06-May-2019]

2015
[122]

Kiwifruit pollination problems,

“Kiwifruit pollination problems,” Science Learning Hub – Pokapū Akoranga Pūtaiao, 2012. [Online]. Available: www.sciencelearn.org.nz/resources/72-kiwifruit-pollination-problems. [Accessed: 19-Mar-2019]

2012
[123]

RoboBee,

“RoboBee,” Science Learning Hub – Pokapū Akoranga Pūtaiao, 2012. [Online]. Available: www.sciencelearn.org.nz/videos/17-robobee. [Accessed: 19-Mar-2019]

2012
[124]

Flight of the robobees,

R. Wood, R. Nagpal, and G.-Y . Wei, “Flight of the robobees,” Sci. Am., vol. 308, no. 3, pp. 60–65, 2013

2013
[125]

Controlled Flight of a Biologically Inspired, Insect-Scale Robot,

K. Y . Ma, P. Chirarattananon, S. B. Fuller, and R. J. Wood, “Controlled Flight of a Biologically Inspired, Insect-Scale Robot,” Science (80-. )., vol. 340, no. 6132, pp. 603–607, 2013

2013
[126]

bvlc_googlenet

“bvlc_googlenet.” [Online]. Available: github.com/BVLC/caffe/tree/rc3/models/bvlc_googlenet. [Accessed: 26-Mar-2019]

2019
[127]

Materially Engineered Artificial Pollinators,

S. A. Chechetka, Y . Yu, M. Tange, and E. Miyako, “Materially Engineered Artificial Pollinators,” Chem, vol. 2, no. 2, pp. 224–239, 2017

2017
[128]

Sticky Solution Provides Grip for the First Robotic Pollinator,

G. J. Amador and D. L. Hu, “Sticky Solution Provides Grip for the First Robotic Pollinator,” Chem, vol. 2, no. 2, pp. 162–164, 2017. 288

2017
[129]

Robotic bees for crop pollination: Why drones cannot replace biodiversity,

S. G. Potts, P. Neumann, B. Vaissière, and N. J. Vereecken, “Robotic bees for crop pollination: Why drones cannot replace biodiversity,” Sci. Total Environ., vol. 642, pp. 665– 667, 2018

2018
[130]

Summon the bee bots: can flying robots save our crops?,

C. Williams, “Summon the bee bots: can flying robots save our crops?,” New Sci., vol. 220, no. 2943, pp. 42–45, 2013

2013
[131]

Design of an Autonomous Precision Pollination Robot,

N. Ohi et al., “Design of an Autonomous Precision Pollination Robot,” in 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2018, pp. 7711–7718

2018
[132]

Farming Robot Changes the Future of Plant Pollination,

T. Peckover, “Farming Robot Changes the Future of Plant Pollination,” 2018. [Online]. Available: https://www.clearpathrobotics.com/2018/05/farming-robot-plant-pollination/? utm_source=newsletter&utm_medium=email&utm_campaign=may. [Accessed: 01-Jun- 2018]

2018
[133]

Available: https://lenses.kowa-usa.com/hc-series/472-lm8hc.html

“LM8HC.” [Online]. Available: https://lenses.kowa-usa.com/hc-series/472-lm8hc.html. [Accessed: 02-Jun-2019]

2019
[134]

Faster {R-CNN:} Towards Real-Time Object Detection with Region Proposal Networks,

S. Ren, K. He, R. B. Girshick, and J. Sun, “Faster {R-CNN:} Towards Real-Time Object Detection with Region Proposal Networks,” CoRR, vol. abs/1506.0, 2015

2015
[135]

Very Deep Convolutional Networks for Large-Scale Image Recognition,

K. Simonyan and A. Zisserman, “Very Deep Convolutional Networks for Large-Scale Image Recognition,” CoRR, vol. abs/1409.1, Sep. 2014

2014
[136]

PASCAL VOC2007 Example Images

“PASCAL VOC2007 Example Images.” [Online]. Available: http://host.robots.ox.ac.uk/pascal/VOC/voc2007/examples/index.html. [Accessed: 07-Jun- 2017]

2017
[137]

Visualizing and Understanding Convolutional Networks,

M. D. Zeiler and R. Fergus, “Visualizing and Understanding Convolutional Networks,” CoRR, vol. abs/1311.2901, 2013

2013 arXiv
[138]

Vision Meets Robotics: The KITTI Dataset,

A. Geiger, P. Lenz, C. Stiller, and R. Urtasun, “Vision Meets Robotics: The KITTI Dataset,” Int. J. Rob. Res., vol. 32, no. 11, pp. 1231–1237, Sep. 2013

2013
[139]

Are we ready for autonomous driving? The KITTI vision benchmark suite,

A. Geiger, P. Lenz, and R. Urtasun, “Are we ready for autonomous driving? The KITTI vision benchmark suite,” in 2012 IEEE Conference on Computer Vision and Pattern Recognition, 2012, pp. 3354–3361

2012
[140]

Development of machine vision and laser radar based autonomous vehicle guidance systems for citrus grove navigation,

V . Subramanian, T. F. Burks, and A. A. Arroyo, “Development of machine vision and laser radar based autonomous vehicle guidance systems for citrus grove navigation,” Comput. Electron. Agric., vol. 53, no. 2, pp. 130–143, 2006

2006
[141]

OpenCV ,

“OpenCV ,” 2019. [Online]. Available: https://opencv.org/. [Accessed: 02-Jun-2019]

2019
[142]

Distance Data Output/UTM-30LX,

“Distance Data Output/UTM-30LX,” 2014. [Online]. Available: www.hokuyo- aut.jp/search/single.php?serial=169. [Accessed: 06-May-2019]

2014
[143]

Intel Core i5-7600K Processor

“Intel Core i5-7600K Processor.” [Online]. Available: https://ark.intel.com/content/www/us/en/ark/products/97144/intel-core-i5-7600k-processor- 6m-cache-up-to-4-20-ghz.html. [Accessed: 02-Jun-2019]. 289

2019
[144]

[Online]

“Puck,” 2019. [Online]. Available: https://velodynelidar.com/vlp-16.html. [Accessed: 02-Jun- 2019]

2019
[145]

Evetar Lens N118B05518W F1.8 f5.5mm 1/1.8

“Evetar Lens N118B05518W F1.8 f5.5mm 1/1.8" - Lenses.” [Online]. Available: www.baslerweb.com/en/products/vision-components/lenses/evetar-lens-n118b05518w-f1-8- f5-5mm-1-1-8/. [Accessed: 02-Jun-2019]

2019
[146]

Male density and arrangement in kiwifruit orchards,

R. Testolin, “Male density and arrangement in kiwifruit orchards,” Sci. Hortic. (Amsterdam)., vol. 48, no. 1, pp. 41–50, 1991

1991
[147]

Image Segmentation Using DIGITS 5,

G. Heinrich, “Image Segmentation Using DIGITS 5,” 2016. [Online]. Available: https://devblogs.nvidia.com/parallelforall/image-segmentation-using-digits-5/. [Accessed: 16-Mar-2017]

2016
[148]

Kiwifruit fruit size distributions,

K. J. McAneney, A. C. Richardson, and A. E. Green, “Kiwifruit fruit size distributions,” New Zeal. J. Crop Hortic. Sci., vol. 17, no. 3, pp. 297–299, 1989

1989
[149]

Automated Pollination of Kiwifruit Flowers,

M. Seabright, J. Barnett, M. H. Jones, P. Martinsen, P. Schaare, J. Bell, H. Williams, M. Nejati, H. A. Seok, J. Lim, A. Scarfe, M. Duke, and B. MacDonald, “Automated Pollination of Kiwifruit Flowers,” in 7th Asian-Australasian Conference on Precision Agriculture (7ACP A), 2017

2017
[150]

Autonomous pollination of individual kiwifruit flowers: Toward a robotic kiwifruit pollinator,

H. Williams, M. Nejati, S. Hussein, N. Penhall, J. Y . Lim, M. H. Jones, J. Bell, H. S. Ahn, S. Bradley, P. Schaare, P. Martinsen, M. Alomar, P. Patel, M. Seabright, M. Duke, A. Scarfe, and B. MacDonald, “Autonomous pollination of individual kiwifruit flowers: Toward a robotic...

2019
[151]

D. I. Jackson, Temperate and Subtropical Fruit Production. Wellington: Butterworths Horticultural Books, 1999

1999
[152]

R. E. Paull and O. Duarte, Tropical Fruits, Volume 2. Oxfordshire: CABI Publishing, 2012

2012
[153]

Chalmers, Hot Climate Wine Growing, Making and Marketing in Southern Italy and its Application in Australia

K. Chalmers, Hot Climate Wine Growing, Making and Marketing in Southern Italy and its Application in Australia. Melbourne: International Specialised Skills Institute, 2013

2013
[154]

Intrinsic Images by Entropy Minimization,

G. D. Finlayson, M. S. Drew, and C. Lu, “Intrinsic Images by Entropy Minimization,” in Computer Vision - ECCV 2004, 2004, pp. 582–595

2004
[155]

Localization and control of an autonomous orchard vehicle,

G. Bayar, M. Bergerman, A. B. Koku, and E. ilhan Konukseven, “Localization and control of an autonomous orchard vehicle,” Comput. Electron. Agric., vol. 115, pp. 118–128, Jul. 2015

2015
[156]

Development of an Autonomous Navigation System using a Two-dimensional Laser Scanner in an Orchard Application,

O. C. Barawid, A. Mizushima, K. Ishii, and N. Noguchi, “Development of an Autonomous Navigation System using a Two-dimensional Laser Scanner in an Orchard Application,” Biosyst. Eng., vol. 96, no. 2, pp. 139–149, 2007

2007
[157]

Row following in pergola structured orchards,

J. Bell, B. A. MacDonald, and H. S. Ahn, “Row following in pergola structured orchards,” in 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2016, pp. 640–645. 290

2016
[158]

A Low-Cost Navigation Strategy for Yield Estimation in Vineyards,

G. Riggio, C. Fantuzzi, and C. Secchi, “A Low-Cost Navigation Strategy for Yield Estimation in Vineyards,” in 2018 IEEE International Conference on Robotics and Automation (ICRA), 2018, pp. 2200–2205

2018
[159]

Orchard navigation using derivative free Kalman filtering,

S. Hansen, E. Bayramoglu, J. C. Andersen, O. Ravn, N. Andersen, and N. K. Poulsen, “Orchard navigation using derivative free Kalman filtering,” in Proceedings of the 2011 American Control Conference, 2011, pp. 4679–4684

2011
[160]

GPS-free Localisation and Navigation for an Unmanned Ground Vehicle for Yield Forecasting in a Vineyard,

S. Marden and M. Whitty, “GPS-free Localisation and Navigation for an Unmanned Ground Vehicle for Yield Forecasting in a Vineyard,” in Proceedings of the 13th International Conference on Intelligent Autonomous Systems, IAS-13, Springer, Padova, Italy, Workshop on Recent Advanc...

2014
[161]

3D perception for accurate row following: Methodology and results,

J. Zhang, A. Chambers, S. Maeta, M. Bergerman, and S. Singh, “3D perception for accurate row following: Methodology and results,” in 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2013, pp. 5306–5313

2013
[162]

Auto Recognition of Navigation Path for Harvest Robot Based on Machine Vision,

B. He, G. Liu, Y . Ji, Y . Si, and R. Gao, “Auto Recognition of Navigation Path for Harvest Robot Based on Machine Vision,” in Computer and Computing Technologies in Agriculture IV: 4th IFIP TC 12 Conference, CCTA 2010, Nanchang, China, October 22-25, 2010, Selected Papers, Pa...

2010
[163]

Bradski and A

G. Bradski and A. Kaehler, Learning OpenCV: Computer Vision with the OpenCV Library. Sebastopol: O’Reilly Media, 2008

2008
[164]

A New Approach to Visual-Based Sensory System for Navigation into Orange Groves,

J. Torres-Sospedra and P. Nebot, “A New Approach to Visual-Based Sensory System for Navigation into Orange Groves,” Sensors, vol. 11, no. 4, pp. 4086–4103, 2011

2011
[165]

A novel vision based row guidance approach for navigation of agricultural mobile robots in orchards,

M. Sharifi and X. Chen, “A novel vision based row guidance approach for navigation of agricultural mobile robots in orchards,” in 2015 6th International Conference on Automation, Robotics and Applications (ICARA), 2015, pp. 251–255

2015
[166]

On the removal of shadows from images,

G. D. Finlayson, S. D. Hordley, and M. S. Drew, “On the removal of shadows from images,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 28, no. 1, pp. 59–68, Jan. 2006

2006
[167]

Simple shadow removal,

C. Fredembach and G. Finlayson, “Simple shadow removal,” in ICPR ’06: Proceedings of the 18th International Conference on Pattern Recognition (ICPR’06), 2006, pp. 832–835

2006
[168]

Vision-based neural network road and intersection detection and traversal,

T. M. Jochem, D. A. Pomerleau, and C. E. Thorpe, “Vision-based neural network road and intersection detection and traversal,” in Proceedings 1995 IEEE/RSJ International Conference on Intelligent Robots and Systems. Human Robot Interaction and Cooperative Robots, 1995, vol. 3, ...

1995
[169]

Dealing with shadows: Capturing intrinsic scene appearance for image-based outdoor localisation,

P. Corke, R. Paul, W. Churchill, and P. Newman, “Dealing with shadows: Capturing intrinsic scene appearance for image-based outdoor localisation,” in 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2013, pp. 2085–2092

2013
[170]

Automating orchards: a system of autonomous tractors for orchard maintenance

S. J. Moorehead, C. K. Wellington, B. J. Gilmore, and C. Vallespi, “Automating orchards: a system of autonomous tractors for orchard maintenance.” Paper presented at the IEEE/RSJ 291 International Conference on Intelligent Robots and Systems, Workshop on Agricultural Robotics,...

2012
[171]

LiDAR Based Tree and Platform Localisation in Almond Orchards,

G. Jagbrant, J. P. Underwood, J. Nieto, and S. Sukkarieh, “LiDAR Based Tree and Platform Localisation in Almond Orchards,” in Field and Service Robotics: Results of the 9th International Conference, L. Mejias, P. Corke, and J. Roberts, Eds. Cham: Springer International Publish...

2015
[172]

A Pipeline for Trunk Detection in Trellis Structured Apple Orchards,

S. Bargoti, J. P. Underwood, J. I. Nieto, and S. Sukkarieh, “A Pipeline for Trunk Detection in Trellis Structured Apple Orchards,” J. F . Robot., vol. 32, no. 8, pp. 1075–1094, 2015

2015
[173]

Orchard and Tree Mapping and Description Using Stereo Vision and Lidar,

M. Nielsen, D. C. Slaughter, C. Gliever, and S. Upadhyaya, “Orchard and Tree Mapping and Description Using Stereo Vision and Lidar,” in International Conference of Agricultural Engineering, 2012

2012
[174]

Dynamic Accuracy of GPS Receivers in Citrus Orchards,

M. Min, R. Ehsani, and M. Salyani, “Dynamic Accuracy of GPS Receivers in Citrus Orchards,” Appl. Eng. Agric., vol. 24, no. 6, pp. 861–868, 2008

2008
[175]

Use of the Hough Transformation to Detect Lines and Curves in Pictures,

R. O. Duda and P. E. Hart, “Use of the Hough Transformation to Detect Lines and Curves in Pictures,” Commun. ACM, vol. 15, no. 1, pp. 11–15, Jan. 1972

1972
[176]

Crop Row Detection on Tiny Plants With the Pattern Hough Transform,

W. Winterhalter, F. V Fleckenstein, C. Dornhege, and W. Burgard, “Crop Row Detection on Tiny Plants With the Pattern Hough Transform,” IEEE Robot. Autom. Lett., vol. 3, no. 4, pp. 3394–3401, Oct. 2018

2018
[177]

Mapping Orchards for Autonomous Navigation,

J. Zhang, S. Maeta, M. Bergerman, and S. Singh, “Mapping Orchards for Autonomous Navigation,” in 2014 ASABE and CSBE/SCGAB Annual International Meeting, 2014

2014
[178]

Perception-driven navigation: Active visual SLAM for robotic area coverage,

A. Kim and R. M. Eustice, “Perception-driven navigation: Active visual SLAM for robotic area coverage,” in 2013 IEEE International Conference on Robotics and Automation, 2013, pp. 3196–3203

2013
[179]

Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age,

C. Cadena, L. Carlone, H. Carrillo, Y . Latif, D. Scaramuzza, J. Neira, I. Reid, and J. J. Leonard, “Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age,” IEEE Trans. Robot., vol. 32, no. 6, pp. 1309–1332, Dec. 2016

2016
[180]

Simultaneous Localization and Mapping,

C. Stachnis, J. J. Leonard, and S. Thrun, “Simultaneous Localization and Mapping,” in Springer Handbook of Robotics, B. Siciliano and O. Khatib, Eds. Cham: Springer, 2016, pp. 1153–1176

2016
[181]

Advances in Neural Information Processing Systems 1,

D. A. Pomerleau, “Advances in Neural Information Processing Systems 1,” D. S. Touretzky, Ed. San Francisco, CA, USA: Morgan Kaufmann Publishers Inc., 1989, pp. 305–313

1989
[182]

Review of research on agricultural vehicle autonomous guidance,

M. Li, K. Imou, K. Wakabayashi, and S. Yokoyama, “Review of research on agricultural vehicle autonomous guidance,” Int. J. Agric. Biol. Eng., vol. 2, no. 3, pp. 1–16, 2009. 293

2009
[183]

Off-road Obstacle Avoidance Through End-to-end Learning,

Y . LeCun, U. Muller, J. Ben, E. Cosatto, and B. Flepp, “Off-road Obstacle Avoidance Through End-to-end Learning,” in Proceedings of the 18th International Conference on Neural Information Processing Systems, 2005, pp. 739–746

2005
[184]

Learning long-range vision for autonomous off-road driving,

R. Hadsell, P. Sermanet, J. Ben, A. Erkan, M. Scoffier, K. Kavukcuoglu, U. Muller, and Y . LeCun, “Learning long-range vision for autonomous off-road driving,” J. F . Robot., vol. 26, no. 2, pp. 120–144, Feb. 2009

2009
[185]

End to End Learning for Self-Driving Cars,

M. Bojarski, D. Del Testa, D. Dworakowski, B. Firner, B. Flepp, P. Goyal, L. D. Jackel, M. Monfort, U. Muller, J. Zhang, X. Zhang, J. Zhao, and K. Zieba, “End to End Learning for Self-Driving Cars,” CoRR, vol. abs/1604.0, 2016

2016
[186]

End-to-end learning for autonomous driving,

U. Muller, “End-to-end learning for autonomous driving,” in O’Reilly AI Conference 2016, 2016

2016
[187]

Vector-based navigation using grid-like representations in artificial agents,

A. Banino et al., “Vector-based navigation using grid-like representations in artificial agents,” Nature, vol. 557, no. 7705, pp. 429–433, 2018

2018
[188]

Automatic Laser Calibration, Mapping, and Localization for Autonomous Vehicles,

J. Levinson, “Automatic Laser Calibration, Mapping, and Localization for Autonomous Vehicles,” Stanford University, 2011

2011
[189]

Automatic Road Detection and Centerline Extraction via Cascaded End-to-End Convolutional Neural Network,

G. Cheng, Y . Wang, S. Xu, H. Wang, S. Xiang, and C. Pan, “Automatic Road Detection and Centerline Extraction via Cascaded End-to-End Convolutional Neural Network,” IEEE Trans. Geosci. Remote Sens., vol. 55, no. 6, pp. 3322–3337, Jun. 2017

2017
[190]

RI Seminar: Ryan Eustice : University of Michigan’s Work Toward Autonomous Cars,

R. M. Eustice, “RI Seminar: Ryan Eustice : University of Michigan’s Work Toward Autonomous Cars,” cmurobotics, 2015. [Online]. Available: www.youtube.com/watch? v=cEtpoeFZxCw. [Accessed: 10-May-2019]

2015
[191]

Making Bertha,

J. Dickmann, N. Appenrodt, and C. Brenk, “Making Bertha,” IEEE Spectr., vol. 51, no. 8, pp. 44–49, 2014

2014
[192]

Radar contribution to highly automated driving,

J. Dickmann, N. Appenrodt, H. Bloecher, C. Brenk, T. Hackbarth, M. Hahn, J. Klappstein, M. Muntzinger, and A. Sailer, “Radar contribution to highly automated driving,” in 2014 44th European Microwave Conference, 2014, pp. 1715–1718

2014
[193]

Making Bertha See,

U. Franke, D. Pfeiffer, C. Rabe, C. Knoeppel, M. Enzweiler, F. Stein, and R. G. Herrtwich, “Making Bertha See,” in 2013 IEEE International Conference on Computer Vision Workshops, 2013, pp. 214–221

2013
[194]

How Cars Learned to See,

U. Franke and S. Gehrig, “How Cars Learned to See,” in Photogrammetric Week 2013, 2013

2013
[195]

Autonomous Driving in Traffic: Boss and the Urban Challenge,

C. Urmson, C. Baker, J. Dolan, P. Rybski, B. Salesky, W. Whittaker, D. Ferguson, and M. Darms, “Autonomous Driving in Traffic: Boss and the Urban Challenge,” AI Mag., vol. 30, no. 2, pp. 17–28, 2009

2009
[196]

Multi-label Point Cloud Annotation by Selection of Sparse Control Points,

R. Monica, J. Aleotti, M. Zillich, and M. Vincze, “Multi-label Point Cloud Annotation by Selection of Sparse Control Points,” in 2017 International Conference on 3D Vision (3DV), 2017, pp. 301–308

2017
[197]

A review of autonomous navigation systems in agricultural environments,

N. Shalal, T. Low, C. McCarthy, and N. Hancock, “A review of autonomous navigation systems in agricultural environments,” in SEAg 2013: Innovative Agricultural Technologies for a Sustainable Future, 2013

2013
[198]

A technical review on navigation systems of agricultural autonomous off- road vehicles,

H. Mousazadeh, “A technical review on navigation systems of agricultural autonomous off- road vehicles,” J. Terramechanics, vol. 50, no. 3, pp. 211–232, 2013

2013
[199]

Robot Farmers: Autonomous Orchard Vehicles Help Tree Fruit Production,

M. Bergerman, S. M. Maeta, J. Zhang, G. M. Freitas, B. Hamner, S. Singh, and G. Kantor, “Robot Farmers: Autonomous Orchard Vehicles Help Tree Fruit Production,” IEEE Robot. Autom. Mag., vol. 22, no. 1, pp. 54–63, Mar. 2015

2015
[200]

A practical obstacle detection system for autonomous orchard vehicles,

G. Freitas, B. Hamner, M. Bergerman, and S. Singh, “A practical obstacle detection system for autonomous orchard vehicles,” in 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2012, pp. 3391–3398

2012
[201]

NEO-M8 series

“NEO-M8 series.” [Online]. Available: www.u-blox.com/en/product/neo-m8-series. [Accessed: 02-Jun-2019]

2019
[202]

OmniSTAR 5120VBS User Manual,

“OmniSTAR 5120VBS User Manual,” 2008. [Online]. Available: www.omnistar.com/Portals/0/downloads/receivers/omnistar/5000 series/5120VBS/5120VBS- doc-user_manual.pdf. [Accessed: 02-Jun-2019]

2008
[203]

2D LiDAR sensors LMS1xx / Outdoor

“2D LiDAR sensors LMS1xx / Outdoor.” [Online]. Available: www.sick.com/nz/en/detection-and-ranging-solutions/2d-lidar-sensors/lms1xx/lms111- 10100/p/p109842. [Accessed: 02-Jun-2019]

2019
[204]

Chameleon3 USB3

“Chameleon3 USB3.” [Online]. Available: www.flir.com.au/products/chameleon3-usb3/? model=CM3-U3-13S2C-CS. [Accessed: 02-Jun-2019]

2019
[205]

Amherst: Adept Technology, 2011

Pioneer 3-AT. Amherst: Adept Technology, 2011

2011
[206]

Gmapping

B. Gerkey, “Gmapping.” [Online]. Available: http://wiki.ros.org/gmapping. [Accessed: 30- Jul-2017]

2017
[207]

Improved Techniques for Grid Mapping With Rao-Blackwellized Particle Filters,

G. Grisetti, C. Stachniss, and W. Burgard, “Improved Techniques for Grid Mapping With Rao-Blackwellized Particle Filters,” IEEE Trans. Robot., vol. 23, no. 1, pp. 34–46, Feb. 2007

2007
[208]

KLD-Sampling: Adaptive Particle Filters and Mobile Robot Localization,

D. Fox, “KLD-Sampling: Adaptive Particle Filters and Mobile Robot Localization,” in In Advances in Neural Information Processing Systems (NIPS), 2001

2001
[209]

Gerkey, “AMCL,” 2018

B. Gerkey, “AMCL,” 2018. [Online]. Available: wiki.ros.org/amcl. [Accessed: 15-May- 2019]

2018
[210]

Robots in Agriculture, Robotics and Autonomous Systems – Vision Challenges and Actions at Royal Society

P. Corke, “Robots in Agriculture, Robotics and Autonomous Systems – Vision Challenges and Actions at Royal Society.” The Royal Society, London, 2015

2015
[211]

A review of defences against common cause failures in reactor protection systems,

M. Kumar, A. Kabra, G. Karmakar, and P. P. Marathe, “A review of defences against common cause failures in reactor protection systems,” in 2015 4th International Conference 294 on Reliability, Infocom Technologies and Optimization (ICRITO) (Trends and Future Directions), 2015, pp. 1–6

2015
[212]

ImageNet: A large-scale hierarchical image database,

J. Deng, W. Dong, R. Socher, L. Li, K. Li, and L. Fei-Fei, “ImageNet: A large-scale hierarchical image database,” in 2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009, pp. 248–255

2009
[213]

Caffe Model Zoo

“Caffe Model Zoo.” [Online]. Available: http://caffe.berkeleyvision.org/model_zoo.html. [Accessed: 08-May-2017]

2017
[214]

caffe/models/bvlc_googlenet,

“caffe/models/bvlc_googlenet,” 2019. [Online]. Available: https://github.com/BVLC/caffe/tree/master/models/bvlc_googlenet. [Accessed: 21-May- 2019]

2019
[215]

caffe/models/bvlc_reference_caffenet,

“caffe/models/bvlc_reference_caffenet,” 2019. [Online]. Available: https://github.com/BVLC/caffe/tree/master/models/bvlc_reference_caffenet. [Accessed: 21- May-2019]

2019
[216]

Row Following in Pergola Structured Orchards by a Monocular Camera Using a Fully Convolutional Neural Network,

J. Bell, B. A. MacDonald, and H. S. Ahn, “Row Following in Pergola Structured Orchards by a Monocular Camera Using a Fully Convolutional Neural Network,” in Australasian Conference on Robotics and Automation (ACRA 2017), 2017, pp. 133–140

2017
[217]

Nvidia Professional Graphics Solutions,

Nvidia, “Nvidia Professional Graphics Solutions,” 2018. [Online]. Available: www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/documents/quadro- mobile-line-card-n18-11x8.5-r4-hr.pdf. [Accessed: 02-Jun-2019]

2018
[218]

Total Stations

“Total Stations.” [Online]. Available: www.trimble.com/Survey/Total-Stations.aspx. [Accessed: 08-May-2017]

2017
[219]

An Analysis of Automated Guided Vehicle Standards to Inform the Development of Mobile Orchard Robots,

J. Bell, B. A. MacDonald, H. S. Ahn, and A. J. Scarfe, “An Analysis of Automated Guided Vehicle Standards to Inform the Development of Mobile Orchard Robots,” IF AC- PapersOnLine, vol. 49, no. 16, pp. 475–480, 2016

2016
[220]

International Organization for Standardization, 2018

ISO 18497:2018 Agricultural machinery and tractors - Safety of highly automated agricultural machines - Principles for design. International Organization for Standardization, 2018

2018
[221]

Safe and reliable: Further development of a field robot,

H. W. Griepentrog, N. A. Andersen, J. C. Andersen, M. Blanke, O. Heinemann, T. Madsen, J. Nielsen, S. M. Pedersen, O. Ravn, and D. Wulfsohn, “Safe and reliable: Further development of a field robot,” Precis. Agric. 2009 - Pap. Present. 7th Eur. Conf. Precis. Agric. ECP A 2009, 2009

2009
[222]

People in the weeds: Pedestrian detection goes off-road,

T. Tabor, Z. Pezzementi, C. Vallespi, and C. Wellington, “People in the weeds: Pedestrian detection goes off-road,” in 2015 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR), 2015, pp. 1–7

2015
[223]

Object tracking: a survey,

A. Yilmaz, O. Javed, and M. Shah, “Object tracking: a survey,” ACM Comput. Surv., vol. 38, pp. 1–45, 2006. 295

2006
[224]

Diversity in Pedestrian Safety for Industrial Environments Using 3D Lidar Sensors and Neural Networks,

J. Bell, B. A. MacDonald, and H. S. Ahn, “Diversity in Pedestrian Safety for Industrial Environments Using 3D Lidar Sensors and Neural Networks,” in 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2018, pp. 7743–7749

2018
[225]

Pedestrian Detection and Tracking Using Three-dimensional LADAR Data,

L. E. Navarro-Serment, C. Mertz, and M. Hebert, “Pedestrian Detection and Tracking Using Three-dimensional LADAR Data,” Int. J. Rob. Res., vol. 29, no. 12, pp. 1516–1528, Oct. 2010

2010
[226]

LADAR-based Pedestrian Detection and Tracking,

L. E. Navarro-Serment, C. Mertz, N. Vandapel, and M. Hebert, “LADAR-based Pedestrian Detection and Tracking,” in 1st Workshop on Human Detection from Mobile Robot Platforms, IEEE ICRA 2008, 2008

2008
[227]

Vehicle Detection from 3D Lidar Using Fully Convolutional Network,

B. Li, T. Zhang, and T. Xia, “Vehicle Detection from 3D Lidar Using Fully Convolutional Network,” in Robotics: Science and Systems XII, 2016

2016
[228]

V oxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection,

Y . Zhou and O. Tuzel, “V oxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection,” CoRR, vol. abs/1711.0, 2017

2017
[229]

3D Object Detection Evaluation 2017

A. Geiger, P. Lenz, C. Stiller, and R. Urtasun, “3D Object Detection Evaluation 2017.” [Online]. Available: www.cvlibs.net/datasets/kitti/eval_object.php?obj_benchmark=3d. [Accessed: 10-Jul-2018]

2017
[230]

However, recent advances such as Hindsight Experience Replay [231] allow some deep reinforcement learning methods to train much quicker than previously possible

and so might be impractically slow to train in a real world application without the benefit of many parallel systems. However, recent advances such as Hindsight Experience Replay [231] allow some deep reinforcement learning methods to train much quicker than previously possibl...
[231]

Multi-View 3D Object Detection Network for Autonomous Driving,

X. Chen, H. Ma, J. Wan, B. Li, and T. Xia, “Multi-View 3D Object Detection Network for Autonomous Driving,” CoRR, vol. abs/1611.07759, 2016

2016 arXiv
[232]

End-to-End Tracking and Semantic Segmentation Using Recurrent Neural Networks,

P. Ondruska, J. Dequaire, D. Z. Wang, and I. Posner, “End-to-End Tracking and Semantic Segmentation Using Recurrent Neural Networks,” CoRR, vol. abs/1604.0, 2016

2016
[233]

Deep tracking in the wild: End- to-end tracking using recurrent neural networks,

J. Dequaire, P. Ondrúška, D. Rao, D. Wang, and I. Posner, “Deep tracking in the wild: End- to-end tracking using recurrent neural networks,” Int. J. Rob. Res., vol. 37, no. 4–5, pp. 492– 512, Jun. 2017

2017
[234]

A Customized Vision System for Tracking Humans Wearing Reflective Safety Clothing from Industrial Vehicles and Machinery,

R. Mosberger, H. Andreasson, and A. J. Lilienthal, “A Customized Vision System for Tracking Humans Wearing Reflective Safety Clothing from Industrial Vehicles and Machinery,” Sensors (Basel)., vol. 14, no. 10, pp. 17952–17980, Oct. 2014

2014
[235]

RI Seminar: John Leonard : Mapping, Localization, and Self-Driving Vehicles,

J. J. Leonard, “RI Seminar: John Leonard : Mapping, Localization, and Self-Driving Vehicles,” cmurobotics, 2015. [Online]. Available: www.youtube.com/watch? v=x5CZmlaMNCs. [Accessed: 24-May-2019]

2015
[236]

ANSI/ITSDF B56.5 - 2012, Safety Standard for Driverless, Automatic Guided Industrial Vehicles and Automated Functions of Manned Industrial Vehicles

ITSDF, “ANSI/ITSDF B56.5 - 2012, Safety Standard for Driverless, Automatic Guided Industrial Vehicles and Automated Functions of Manned Industrial Vehicles.” Industrial Truck Standards Development Foundation, Washington DC, 2012

2012
[237]

2015 ITS America Annual Meeting Opening Plenary Keynote,

C. Urmson, “2015 ITS America Annual Meeting Opening Plenary Keynote,” ITS America,

2015
[238]

New Mobility World Executive Forum Part 3 Chris Urmson,

C. Urmson, “New Mobility World Executive Forum Part 3 Chris Urmson,” New Mobility World Startup Zone, 2015. [Online]. Available: www.youtube.com/watch?v=_-sil7Ammck. [Accessed: 05-Nov-2015]. 296

2015
[240]

Dynamic Routing Between Capsules,

S. Sabour, N. Frosst, and G. E. Hinton, “Dynamic Routing Between Capsules,” CoRR, vol. abs/1710.0, 2017

2017
[241]

Matrix capsules with {EM} routing,

G. E. Hinton, S. Sabour, and N. Frosst, “Matrix capsules with {EM} routing,” in International Conference on Learning Representations, 2018

2018
[242]

Improved Explainability of Capsule Networks: Relevance Path by Agreement,

A. Shahroudnejad, A. Mohammadi, and K. N. Plataniotis, “Improved Explainability of Capsule Networks: Relevance Path by Agreement,” CoRR, vol. abs/1802.10204, 2018

2018 arXiv
[243]

Object Detection Evaluation 2012

A. Geiger, P. Lenz, C. Stiller, and R. Urtasun, “Object Detection Evaluation 2012.” [Online]. Available: www.cvlibs.net/datasets/kitti/eval_object.php. [Accessed: 02-Jun-2019]

2012
[244]

Learning Hand-Eye Coordination for Robotic Grasping with Deep Learning and Large-Scale Data Collection,

S. Levine, P. Pastor, A. Krizhevsky, and D. Quillen, “Learning Hand-Eye Coordination for Robotic Grasping with Deep Learning and Large-Scale Data Collection,” CoRR, vol. abs/1603.0, 2016

2016
[245]

Neuroscience-Inspired Artificial Intelligence,

D. Hassabis, D. Kumaran, C. Summerfield, and M. Botvinick, “Neuroscience-Inspired Artificial Intelligence,” Neuron, vol. 95, no. 2, pp. 245–258, 2017

2017
[246]

Hindsight Experience Replay,

M. Andrychowicz, F. Wolski, A. Ray, J. Schneider, R. Fong, P. Welinder, B. McGrew, J. Tobin, P. Abbeel, and W. Zaremba, “Hindsight Experience Replay,” in Advances in Neural Information Processing Systems 30 (NIPS 2017), 2017

2017
[247]

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks,

C. Finn, P. Abbeel, and S. Levine, “Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks,” in Proceedings of the 34th International Conference on Machine Learning, 2017, vol. 70, pp. 1126–1135

2017
[248]

Emergence of Locomotion Behaviours in Rich Environments,

N. Heess, D. TB, S. Sriram, J. Lemmon, J. Merel, G. Wayne, Y . Tassa, T. Erez, Z. Wang, A. Eslami, M. Riedmiller, and D. Silver, “Emergence of Locomotion Behaviours in Rich Environments,” CoRR, vol. abs/1707.0, 2017

2017
[249]

3D Simulation for Robot Arm Control with Deep Q-Learning,

S. James and E. Johns, “3D Simulation for Robot Arm Control with Deep Q-Learning,” CoRR, vol. abs/1609.0, 2016

2016
[250]

Transferring End-to-End Visuomotor Control from Simulation to Real World for a Multi-Stage Task,

S. James, A. J. Davison, and E. Johns, “Transferring End-to-End Visuomotor Control from Simulation to Real World for a Multi-Stage Task,” in Proceedings of the 1st Annual Conference on Robot Learning, 2017, vol. 78, pp. 334–343

2017
[251]

Sim-to-Real Robot Learning from Pixels with Progressive Nets,

A. A. Rusu, M. Večerík, T. Rothörl, N. Heess, R. Pascanu, and R. Hadsell, “Sim-to-Real Robot Learning from Pixels with Progressive Nets,” in Proceedings of the 1st Annual Conference on Robot Learning, 2017, vol. 78, pp. 262–270

2017
[252]

Deep Reinforcement Learning for Dexterous Manipulation with Concept Networks,

A. Gudimella, R. Story, M. Shaker, R. Kong, M. Brown, V . Shnayder, and M. Campos, “Deep Reinforcement Learning for Dexterous Manipulation with Concept Networks,” CoRR, vol. abs/1709.06977, 2017. 297

2017 arXiv
[253]

Overcoming Exploration in Reinforcement Learning with Demonstrations,

A. Nair, B. McGrew, M. Andrychowicz, W. Zaremba, and P. Abbeel, “Overcoming Exploration in Reinforcement Learning with Demonstrations,” CoRR, vol. abs/1709.10089, 2017. 298

2017 arXiv
[2015]

Available: www.youtube.com/watch?v=9m0xMeWONhs

[Online]. Available: www.youtube.com/watch?v=9m0xMeWONhs. [Accessed: 05- Nov-2015]

2015

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

Then the overlaps between calyx instances and skin instances were found

In order to find calyx-skin pairs, the calyx segmentation outputs were dilated. Then the overlaps between calyx instances and skin instances were found. For each calyx instance, the skin instance with the maximum overlap was assigned to the calyx instance

[2] [2]

The centre of each calyx instance was found using the median x-coordinate and y-coordinate in the calyx segmentation output. 273

[3] [3]

For each calyx-skin pair, Principal Components Analysis (PCA) of the skin instance points was used to find the centre of the skin and the directions of the principal components

[4] [4]

It was thought that this procedure might work well for unoccluded fruit

For each calyx-skin pair, the calyx centre was reflected about the second principal component. It was thought that this procedure might work well for unoccluded fruit. However, for occluded fruit it was postulated that PCA might not find the principal components of the fruit accurately. Although, for occluded fruit, it was thought that the collision detec...

[5] [5]

turn left

Repeating steps 1-4 multiple times. The camera images collected were post-processed into 2 classes for classification; “turn left” and “turn right”. The resulting data was used to train an AlexNet CNN, using the hyperparameters given in Table 53. The resulting average validation accuracy from 3 training runs was 88%. This result 277 seemed quite poor for ...

2016

[6] [6]

A fixed size region of interest around the centre of a detected fruit was set

[7] [7]

The points in the region of interest were transformed to the robot arm coordinate system

[8] [8]

The points were transformed into the coordinate system of the gripper, for the pose of the gripper at the instant of fruit detachment