SOFTWARE-HARDWARE COMPLEX FOR OBSTACLE SEGMENTATION WITH U-NET ARCHITECTURE FOR AUTONOMOUS AGRICULTURAL MACHINERY
Abstract
Agriculture plays a fundamental role in ensuring food security and meeting the population's
needs for food products. Optimization of agricultural crop production and increasing efficiency
are essential tasks for modern agriculture. In this regard, more attention is being given to the
development and implementation of autonomous agricultural systems capable of automating and
optimizing various production processes. However, the effectiveness of autonomous systems is
limited by the insufficient development of obstacle detection systems and decision-making algorithms.
When agricultural machinery and other autonomous vehicles encounter obstacles in their
path, precise and rapid recognition of these obstacles plays a decisive role in making appropriate
decisions to avoid accidents. This article presents a software-hardware complex for obstacle segmentation using the U-Net architecture, designed to overcome these limitations in autonomous
agricultural systems. The U-Net architecture is renowned for its ability to accurately recognize
objects in images, making it an attractive choice for machine vision systems in agricultural conditions.
The presented complex boasts high performance and enables real-time obstacle segmentation,
including columns, trees, and shrubbery, during the movement of agricultural machinery
along a designated trajectory. This ensures precise decision-making and avoidance of accidents,
significantly enhancing the efficiency and safety of autonomous systems in agricultural production.
Field tests have confirmed the effectiveness and applicability of the proposed solutions under
real agricultural conditions. The presented software-hardware complex with U-Net architecture
opens up new possibilities for autonomous agricultural technology, promoting increased productivity
and efficiency in agriculture. It represents a significant step in the development of modern
agricultural technologies and contributes to the use of autonomous systems to enhance agricultural
production and improve productivity.
References
tekhnologicheskikh protsessov v sel'skom khozyaystve [Neural network modeling of technological
processes in agriculture], Sovremennoe sostoyanie, problemy i perspektivy razvitiya
mekhanizatsii i tekhnicheskogo servisa agropromyshlennogo kompleksa: Mater.
mezhdunarodnoy nauchno-prakticheskoy konferentsii Instituta mekhanizatsii i tekhnicheskogo
servisa, Kazan', 07–08 iyunya 2019 g. [Current state, problems and prospects for the development
of mechanization and technical service of the agro-industrial complex: Proceedings of
the international scientific and practical conference of the Institute of mechanization and technical
service, Kazan, June 07-08, 2019]. Kazan': Kazanskiy gosudarstvennyy agrarnyy
universitet, 2019, pp. 182-184. EDN WOJDMP.
2. Chikrin D.E. Universal'nye komplekty avtomatizatsii sel'skokhozyaystvennoy tekhniki dlya
robotizirovannykh agrokhozyaystv [Universal kits for automation of agricultural machinery
for robotic farms], Mezhdunarodnaya agropromyshlennaya vystavka «AgroVolga 2021» [International
agro-industrial exhibition "AgroVolga 2021"], 2021.
3. Tormagov T.A., Generalov A.A., Shavin M.Yu., Rapoport L.B. Zadachi upravleniya
dvizheniem avtonomnykh kolesnykh robotov v tochnom zemledelii [Problems of motion control
of autonomous wheeled robots in precision agriculture], Giroskopiya i navigatsiya
[Gyroscopy and navigation], 2022, Vol. 30, No. 1 (116), pp. 39-60. DOI 10.17285/0869-
7035.0083. EDN GWCIQX.
4. Sabirov R.F., Valiev A.R., Medvedev V.M., Shafigullin G.T., Galiullin I.G. Svid. o gos. registratsii
progr. dlya EVM 2021667661 Ros. Federatsiya. Programmnyy modul' segmentatsii ob"ektov
prepyatstviy derevo i stolb na osnove neyronnykh setey [Certificate of state registration of computer
programs 2021667661 Russian Federation. Software module for segmentation of obstacle objects
tree and pillar based on neural networks], Appl. 10/20/2021; registered 11/01/2021; publ.
01.11.2021, Bull. No. 11]. Available at: https://www.elibrary.ru/item.asp?id=47257657
5. Gonsales R., Vuds R. Tsifrovaya obrabotka izobrazheniy [Digital image processing]. Moscow:
Tekhnosfera, 2005, 1072 p.
6. Sikorskiy O.S. Obzor svertochnykh neyronnykh setey dlya zadachi klassifikatsii izobrazheniy
[Review of convolutional neural networks for the problem of image classification], Novye
informatsionnye tekhnologii v avtomatizirovannykh sistemakh [New information technologies
in automated systems], 2017, No. 20, pp. 37-42.
7. Nikolenko S., Kadurin A., Arkhangel'skaya E. Glubokoe obuchenie [Deep learning].
St. Petersburg: Piter, 2018, 480 p.
8. Atole R.R., Park D. A Multiclass Deep Convolutional Neural Network Classifier for Detection
of Common Rice Plant Anomalies, (IJACSA) International Journal of Advanced Computer
Science and Applications, 2018, Vol. 9, No. 1, pp. 67-70.
9. Rajmohan R., Pajany M., Rajesh R., Raghu Raman D., Prabu U. Smart paddy crop disease
identification and management using deep convolution neural network and SVM classifier, International
Journal of Pure and Applied Mathematics, 2018, Vol. 118, No. 15, pp. 255-264.
10. Athanikar G., Badar P. Potato Leaf Diseases Detection and Classification System, International
Journal of Computer Science and Mobile Computing, 2016, Vol. 5, Issue 2, pp. 76-88.
11. Sladojevic S., Arsenovic M., Anderla A., Culibrk D., and Stefanovic D. Deep Neural Networks
Based Recognition of Plant Diseases by Leaf Image Classification, Computational Intelligence
and Neuroscience, 2016, 11 p.
12. Abdullahi H.S., Zubair O.M. Advances of image processing in Precision Agriculture: Using
deep learning convolution neural network for soil nutrient classification, Journal of Multidisciplinary
Engineering Science and Technology (JMEST), August 2017, Vol. 4, Issue 8,
pp 7981-7987.
13. Huang H., Deng J., Lan Y., Yang A., Deng X., Zhang L. A fully convolutional network for
weed mapping of unmanned aerial vehicle (UAV) imagery // PLoS ONE 13(4): e0196302.
– 2018.
14. Inkyu Sa, Zetao Chen, Marija Popovic, Raghav Khanna, Frank Liebisch, Juan Nieto, Roland
Siegwa. WeedNet: Dense Semantic Weed Classification Using Multispectral Images and MAV
for Smart Farming, IEEE Robotics and Automation Letters, 2018, Vol. 3 (1), pp. 588-595.
15. Potena C., Nardi D., Pretto A. Fast and Accurate Crop and Weed Identification with Summarized
Train Sets for Precision Agriculture, IAS 2016: Intelligent Autonomous Systems, 2017,
14, pp 105-121.
16. Yao C., Zhang Y., Zhang Y., and Liu H. Application of convolutional neural network in classification
of high resolution agricultural remote sensing images, Int. Arch. Photogramm. Remote
Sens. Spatial Inf. Sci., XLII2/W7, 2017, pp. 989-992.
17. Atole R.R., Park D. A Multiclass Deep Convolutional Neural Network Classifier for Detection
of Common Rice Plant Anomalies, (IJACSA) International Journal of Advanced Computer
Science and Applications, 2018, Vol. 9, No. 1, pp. 67-70.
18. Ganchenko V., Doudkin A. Image Semantic Segmentation Based on Convolutional Neural
Networks for Monitoring Agricultural Vegetation, Communications in Computer and Information
Science, Springer, 2019, 2019, Part 5, Vol. 1055, pp. 52-63.
19. Adam: A Method for Stochastic Optimization. Available at: https://arxiv.org/abs/1412.6980.
20. SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation.
Available at: https://ieeexplore.ieee.org/document/7803544/.
21. Densely Connected Convolutional Networks. Available at: https://arxiv.org/pdf/
1608.06993.pdf.
22. Chikrin D.E. Metodologicheskie osnovy proektirovaniya infokommunikatsionnykh sistem
avtomobil'nykh transportnykh sredstv vysokoy stepeni avtomatizatsii: dis. … d-ra tekhn. nauk
[Methodological foundations for the design of infocommunication systems for automobile vehicles
of a high degree of automation: dr. of eng. sc. diss.], 05.13.01. Kazan', 2021, 399 p.
23. Egorchev A.A. Verifitsiruemye sistemy virtual'nogo modelirovaniya bespilotnykh
transportnykh sredstv: dis. … kand. tekhn. nauk [Verifiable systems for virtual simulation of
unmanned vehicles: cand. of eng. sc. diss.], 05.13.01. Kazan', 2021, 340 p.
24. Kots M.V., Chukanov V.S. U-Net adaptation for multiple instance learning, Journal of Physics:
Conference Series, Saint Petersburg, 21–22 March 2019. Vol. 1236. Saint Petersburg: Institute
of Physics Publishing, 2019, pp. 012061. DOI: 10.1088/1742-6596/1236/1/012061. EDN
SJUOKU.
25. Ganchenko V.V., Dudkin A.A., Sheleg S.V. Raspoznavanie sel'skokhozyaystvennoy
rastitel'nosti na izobrazheniyakh zemnoy poverkhnosti na osnove svertochnoy neyronnoy seti
U-Net [Recognition of agricultural vegetation on images of the earth's surface based on the
convolutional neural network U-Net], Big Data and Advanced Analytics, 2021, No. 7-1,
pp. 110-116. EDN ZDCYJM.
26. Valiev A.R., Manuel' Binelo, Ziganshin B.G., Sabirov R.F., SHafigullin G.T., Galiullin I.G.
Bespilotnyy traktor [Unmanned tractor], Vestnik NTSBZhD [Vestnik NTsBZhD], 2021, No. 4
(50), pp. 69-75.
27. Galiullin I.G. Sistema avtonomnogo upravleniya dvizheniem mashinno-traktornogo agregata s
ispol'zovaniem otechestvennoy elementnoy bazy [The system of autonomous control of the
movement of the machine-tractor unit using the domestic element base], Izvestiya Kabardino-
Balkarskogo nauchnogo tsentra RAN [Izvestiya of the Kabardino-Balkarian Scientific Center
of the Russian Academy of Sciences], 2022, No. 6 (110), pp. 92-98. DOI: 10.35330/1991-
6639-2022-6-110-92-98. EDN NZIISF.
