PRE2018 4 Group3
Group Members
| Name | Student Id |
|---|---|
| Han Wei Chia | 1002684 |
| Niek Brekelmans | 1017203 |
| Floris Verheijen | 0948592 |
| Insert Name | Insert ID |
| Insert Name | Insert ID |
Problem statement
Objective
User, society and enterprise
Users
Society
There are more honey bees in this world than any other type of bee and pollinating insects. This means that honey bees are the most important pollinators of our food crops. Approximately one third of our food relies on the pollination by bees. Without honey bees, we would have a global food crisis that would kill a lot of people. This food shortage in case of an extinction will be prevented if an artificial pollinator replaces bees in time. The protection of our food chain is essential and vital to humanity's survival.
Enterprise
Plants will be in trouble if pollinators die out. A lot of them would go extinct. This would lead to mass disruption of insect and wildlife life cycles. It would be hard to predict exactly what would happen, but there would be many negative impacts on user and society alike. There will be huge demand for other (Artificial) Pollination solution. Robotic bees could be the solution and be very beneficial for enterprises to invest in
Requirements
The things users will require the drones to meet are;
- The reusability of the drones
- They need to be mass producible
- Environment friendly materials need to be used preferably bio degradable
- The drones need to be energy efficient so they last long on one charge
- Easy to control or automated so that 1 person can control multiple drones at once
- During pollination of the flowers the flowers should not be damaged by the drone
- The drones need to be replaceble by one another like real bees are in a swarm
- There must be a way to locate the drones if they break in order to reuse them
Approach
The following appraoch will be used to meet the requirements:
First a literature study will be done on the techniques and requirements described earlier. Next will be a literature study on the current state of the art of artificial pollination. when the research is done a model and/or prototype will be build.
Milestones
| Week | Milestones |
|---|---|
| 1 |
|
| 2 |
|
| 3 |
|
| 4 |
|
| 5 |
|
| 6 |
|
| 7 |
|
| 8 |
|
| 9 |
|
Deliverables
Planning
Our up-to-date planning can be found with the following link: [1].
State of the Art
Pollination
Artificial Pollination in Kiwifruit and Olive Trees[1]
In this study, they tested what the best way to collect,store and spread pollen for kiwifruits. Pollen samples were collected with two different systems, but was irrelevant to the conclusion. They timing of when and how to store was more important. Th best way to store to guarantee the highest qualtiy of pollen obtained when the pollen were picked up from the collecting machines about every hour. This is to avoid any stres on the pollen. For short term storage the pollen needed to be stored at 4°C for no more than 7 days. For long tern storage the pollen needed to be stored at −18°C for no more than 3 years low humidity or pre-dried to 10–12% with silica gel at 4°C.
For spreading the pollen they used liquid and dry pollination with varying machines in different flowering stages of the kiwifruit flower. There both as effect if done at the specific flowering. for liquid pollination it was Early Petals Fall and for dry pollination it was Petals Fall.
They used the same technique on olive trees to better understand the moment for pollination in relation to the flowering stage during flowering as they were as they were effective as well.
Pollination efficiency of artificial and bee pollination practices in kiwifruit [2]
In this study they state that the efficiency of artificial pollination has never been compared with that provided by bees and will do so themselves. When comparing bee pollination with artificial. Bee pollination did not only increase the number of kiwifruit produced, but also the number of seeds per fruit, fruit weight and even higher homogeneityin.
Something to also note:
Almost all the fruits produced in the bee-pollinated plants were of export quality while that of artificially pollinated were not.This is because Artificially pollination happened once, when ∼10% of all flowers remained as buds.as for the open flower that were sprayed with pollen, some of them were already senescent. The senescent flowers causes higher chances of producing malformed fruits or no fruit at all.
Effects of natural and artificial pollination on fruit and offspring quality [3]
In this study they research the effects natural and artificial pollination on cape gooseberry. The test the effects of fruit and offspring characteristics on honey and bumble bee pollination compared to manual outcrossing and autonomous self-pollination. Compared to manual and self-pollination, bee pollination increased fruit size, seed set and germination rates. On the other hand , plant growth rate and herbivore resistance were significantly and marginally greater in manually outcrossed plants compared to self-pollinated offspring, suggesting that inbreeding reduces offspring quality. Herbivore resistance and plant growth did not differ between one honeybee visit and self-pollination suggesting that multiple pollinator visits are needed to prevent inbreeding events. bees visitation can significantly alter ecologically and economically relevant traits in this agroecosystem.
Materially Engineered Artificial Pollinators [4]
In this study, multifunctionality from synthesized ionic liquidgels (ILGs) for biotechnology is presented. ILGs exhibit unique properties and coating vertically aligned animal hair with ILGs can be used for effective pollen collection. When place onto a radiowave-controllable UAV it could successfully pollinate L. japonicumflowers.
Development of strawberry pollination system using ultrasonic radiation pressure [5]
In this study they developed an artificial pollination system using ultrasonic radiation pressure as a substitute technique for bee pollination for strawberry cultivation in a plant factory. It has a higher marketable rate than that of no pollination treatment or brush pollination.
(Autonomous) Drones
Autonomous drone is making test flights in Kansas, Illinois [6]
In this project, a drone was created which can fly without an operator or pilot on the scene. It has been created for the purpose of surveillance. This project shows how an autonomous drone which keeps track of a map spawns more than 30 GB of data to fly in an area of around 400 hectares.
Watching the watchmen: Drone privacy and the need for oversight [7]
This paper explores the privacy concerns that is associated with drones and other UAVs. It shows how a 'privacy by design (PbD)' approach helps to ensure that the aqcuired data is protected and the privacy is protected from an early stage of development.
Privacy, data protection and ethics for civil drone practice: A survey of industry, regulators and civil society organisations [8]
This article presents the findings from a survey of the drone industry, regulators and civil society organisations. It uses these results to show that the drone industry is diverse in applications and payloads. The industry sometimes has a lack of knowledge about privacy, ethics and data protection. Operators are often not aware of their obligations within the European law about these subjects. Bringing together watchdogs and regulatory organisations could help to educate drone operators and members of the public.
Experimentally Validated Extended Kalman Filter for UAV State Estimation Using Low-Cost Sensors [9]
Visually based velocity and position estimations can make sure an UAV does not depend on GPS systems. This paper explores a sensor-fusion algorithm, which uses a few different sensors to achieve this. In the experiments, varying parameters were removed in case of different environmental situations. The results show that the velocity and attitude can be estimated, dispite various (indoor) environments.
Image recognition
Cats or CAT scans: transfer learning from natural or medical image source datasets? [10]
In this article the usefullness of transfer learning is explaned for medical image analysis. Because in medical image analysis there is not much data avaiilable for training a neural network. To do this there was a large amount of data used that had nothing to do with medical images but that could be classified in different cathegories.
Multispectral images of flowers reveal the adaptive significance of using long-wavelength-sensitive receptors for edge detection in bees[11]
in this article
Deap Learning[12]
in this article
Deep Learning of Representations for Unsupervised and Transfer Learning [13]
Food image recognition using deep convolutional network with pre-training and fine-tuning [14]
References
- ↑ Tacconi Gianni and Michelotti Vania (June 6th 2018). Artificial Pollination in Kiwifruit and Olive Trees, Pollination in Plants, Phatlane William Mokwala, IntechOpen, DOI: 10.5772/intechopen.74831. Available from: https://www.intechopen.com/books/pollination-in-plants/artificial-pollination-in-kiwifruit-and-olive-trees
- ↑ Agustín Sáez, Pedro Negri, Matias Viel, Marcelo A. Aizen, Pollination efficiency of artificial and bee pollination practices in kiwifruit Scientia Horticulturae, Volume 246, 2019, Pages 1017-1021, ISSN 0304-4238, http://doi.org/10.1016/j.scienta.2018.11.072.(http://www.sciencedirect.com/science/article/pii/S0304423818308525)
- ↑ Alexander Chautá-Mellizo, Stuart A. Campbell, Maria Argenis Bonilla, Jennifer S. Thaler, Katja Poveda, Effects of natural and artificial pollination on fruit and offspring quality, Basic and Applied Ecology, Volume 13, Issue 6, 2012, Page 524-532, ISSN 1439-1791, https://doi.org/10.1016/j.baae.2012.08.013. (http://www.sciencedirect.com/science/article/pii/S143917911200093X)
- ↑ Svetlana A. Chechetka, Yue Yu, Masayoshi Tange, Eijiro Miyako, Materially Engineered Artificial Pollinators, Chem, Volume 2, Issue 2, 2017, Pages 224-239, ISSN 2451-9294, https://doi.org/10.1016/j.chempr.2017.01.008. (http://www.sciencedirect.com/science/article/pii/S2451929417300323)
- ↑ Hiroshi Shimizu, Taito Sato, Development of strawberry pollination system using ultrasonic radiation pressure, IFAC-PapersOnLine, Volume 51, Issue 17, 2018, Pages 57-60, ISSN 2405-8963, https://doi.org/10.1016/j.ifacol.2018.08.060 (http://www.sciencedirect.com/science/article/pii/S2405896318311765)
- ↑ P. J. Griekspoor, "Autonomous drone is making test flights in Kansas, Illinois," Southwest Farm Press, 2018. Available: https://search.proquest.com/docview/2088344724?accountid=27128
- ↑ Jenkins, Ben. "Watching the watchmen: Drone privacy and the need for oversight." Ky. LJ 102 (2013): 161.
- ↑ Rachel L. Finn, David Wright, "Privacy, data protection and ethics for civil drone practice: A survey of industry, regulators and civil society organisations", Computer Law & Security Review, vol. 32, num. 4, p.p. 577 - 586
- ↑ Driessen, S. P. H., et al. "Experimentally Validated Extended Kalman Filter for UAV State Estimation using Low-Cost Sensors." IFAC-PapersOnLine, vol. 51, no. 15, 2018, pp. 43-48. SCOPUS, www.scopus.com, doi:10.1016/j.ifacol.2018.09.088.
- ↑ Cheplygina, V. (2019). Cats or CAT scans: Transfer learning from natural or medical image source data sets?.
- ↑ Vasas, V., Hanley, D., Kevan, P. and Chittka, L. (2019). Multispectral images of flowers reveal the adaptive significance of using long-wavelength-sensitive receptors for edge detection in bees.
- ↑ LeCun, Y., Bengio, Y. and Hinton, G. (2015). Deep learning. Nature, 521, p.436.
- ↑ Bengio. Y (2012)Deep Learning of Representations for Unsupervised and Transfer Learning JMLR
- ↑ K. Yanai and Y. Kawano, "Food image recognition using deep convolutional network with pre-training and fine-tuning," 2015 IEEE International Conference on Multimedia & Expo Workshops (ICMEW), Turin, 2015, pp. 1-6. doi: 10.1109/ICMEW.2015.7169816 keywords: {feature extraction;food products;image classification;image recognition;neural nets;food image recognition;deep convolutional neural network;food photo recognition task;fine-grained visual recognition;DCNN-related techniques;large-scale ImageNet data;pre-trained DCNN;fine-tuned DCNN;activation feature extraction;UEC-FOOD100;UEC-FOOD256;food classifier;Twitter photo data;food photo mining;Feature extraction;Accuracy;Twitter;Image recognition;Image color analysis;Data mining;Training;deep convolutional neural network food recognition Twitter photo mining}, URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7169816&isnumber=7169738