Introduction

Artificial Intelligence can be defined as a computer system developed to perform tasks normally requiring a human input. This usually includes decision making, image or speech recognition or assessment of qualitative data. As the 2020s move forward A.I. has become increasingly sophisticated as such become increasingly integrated in various industries. One industry is agriculture, which is arguably one of the most important industries for the world due to it being key to human food production. As human population rises each year more and more pressure is being placed on crop and livestock farmers to not only produce increasing levels of food but also guarantee and good quality of food and welfare of stock. As such new paradigms need to be developed to aid farmers complete these goals while still allowing farmers to earn a decent living and have ample job opportunities. With this in mind strides have been made in the last 3 years to develop A.I. programs to streamline the farming process for the benefit of both the consumer and for farmers globally. This article’s purpose is to assess what A.I. and other linked technology is currently being used in livestock farming alongside it’s their benefits and drawbacks. It will also cover potential areas of development where A.I. could be further developed to further benefit livestock farmers in the future.

Application and benefits if A.I. in livestock farming

Machine learning has already been successfully implemented in the following fields:

• Food security

• Climatology

• Aquaculture

• Robotics

• Economics

• Meteorology

• Medicine

• Bioinformatics

• Biochemistry

Artificial neural networks modelled on human brain functionality can utilise complex layers of data and can help create smart decision making for farmers. Machine learning can aid in monitoring animal behaviour and disease detection as well as create an accurate estimation of economic balances for products. Use of sensors with magneto meters and 3-axis accelerometers can help predict oestrus patterns and dietary changes in livestock. Machine learning could also assist in water management keeping tabs on water quality, monitor for pathogens, predict weather patterns and predict changes in water consumption rates (Liakos. K. G, et al, 2018). A.I. could be used to monitor animal growth rates to adjust feed accordingly with both the animal’s nutritional need and the farmer’s need to produce products to sell. Utilising a variety of sensors animal weight, milk quality, movement, sleep cycles and air quality can be monitored and fed into an algorithm. Due to low statistical chance of error the use of these robotic systems to administer medication and vaccines to animals could help keep livestock healthy (Patel. H, et al, 2022).

A study in 2021 found that dairy farms in EU countries were highly unstable in terms of milk production and quality. Stability of milk production varies greatly from country to country this is a problem for dairy farmers due to demand for milk rapidly on the rise. The use of automated milking systems (AMS) is on the rise to tackle these issues especially in Eastern Europe. This is due to increased quality of harvested milk, welfare of dairy cows and reduction and redirection of labour on said dairy farms. Use of A.I. especially in data analysis can help farms monitor and guarantee a higher level of quality and food security in their products (Micle. D. E, et al, 2021). Integration of A.I. and drone technology could allow for more efficient ongoing stock surveillance. Machine learning could also be used to detect and predict disease outbreaks for both crops and livestock. This kind of tech has been in use since around 2012 in cattle management by use of overhead cameras and vision algorithms. This allows farmers to view cattle in real time and alert farmers as soon as an issue arises. This can also include keeping track of births and signs of illness and strange behaviour as well as track the animal’s food and water availability. Automatic weeding could also be used to not only save farmers time but also keep livestock fields clear of potentially dangerous plants such as ragwort avoiding unnecessary fatalities (Sharma. S, et al, 2022).

Cloud hosted platforms allow for multiple systems to work in tandem with each other allowing for coordinated data streams to give deep learning programs an accurate picture of the farm it is being employed in. One drawback is the systems strong reliance on strong internet signal which can be a problem in rural areas. Herdsman+ is a system being developed and rolled out for dairy farmers to integrate multiple data mining sources into one interface for farmers (Michie. C, et al, 2022). In the face of climate change one of the biggest problems facing livestock farming and transportation is increased ambient temperatures and unpredictable rainfall. In addition, water availability is a problem, in short unpredictable weather as such livestock farming must become less reactive in terms of weather conditions. Another area of livestock farming must evolve in is for it to depend less on invasive tests to monitor livestock welfare such as tissue collection surgically implanted sensors. Biometrics can vary depending on the animal in question due to variations in behavioural norms and health requirements. However, collection of biometrics will mostly involve remote sensors and machine vision to feed data to a machine learning program. However, some sensors will still need to be attached to the animal though this should be minimal (Fuentes. S, et al, 2022).

Smart farming has been developed not only in the sense of efficiency but also for livestock welfare standards in transportation, slaughter and rearing. These welfare standards are categorised as natural life, emotional state and base health and function. In pigs specifically this can be assessed through appearance phenotype e.g., body size and weight, behavioural performance and sound recognition. To overcome human limitations in observing these traits new technologies are implemented such as smart sensors IOTs and general A.I. smart farming will utilise smart sensors to keep track of animal health by monitoring pen layout, temperature and humidity. This then feeds A.I. algorithms through industrial internet servers, the A.I. can then analyse pig data to help farmers with decision making. Cameras and microphones can be used to analyse response characteristics and can suggest strategies to optimise production/ reproduction, avoid over feeding and reduce waste allowing for more sustainable farming. One problem based in deep learning with herd animals is occlusion an over lapping noise in image data with captive specimens as the acquisition object. To get around this a study in 2020 proposed key point detection based on deep convolutional encoder- decoder networks. This determines each animal by a key point heat map with an offset vector field, this proved more effective than traditional bounding box-based detection (Wang. S, et al, 2022).

A study in 2022 shows rapid growth of interest and development for use of artificial intelligence in the dairy farm sector. A.I. can be used to detect disease in cattle such as ketosis and metritis as shown by a study in 2016 this system can detect transmission of disease earlier than humans. This allows for corrective action to be taken sooner and can limit the effect of disease, cows can be monitored by sensors and rumination collars. These can be connected to algorithms that can keep track of weight, milk yield and food intake factors which can be indicators of disease. Machine learning can also be used to keep track of behaviours which can indicate periods of oestrus and determine optimal times for insemination. Similar metrics can also be used to predict when calving is most likely to occur and also help farmers select optimal types of semen for use in insemination. This can be done by analysing health records of possible males and the health of their existing offspring to work out the chances of currant insemination will produce healthy offspring. By keeping taps on previous milk yield as well as individual health A.I. can help estimate future milk yields helping farmers make better decisions this includes selecting high quality cows for breeding, culling bad quality cows, nutritional regimes and regrouping. Keeping track of weight levels, feed intake and production can help corrective herd health and developing efficient strategies. Several dairy farms have automated the milking process such as laser guided robots to detect teats and extract milk. As well as automated systems to deliver feed and extract waste from their pens (De Vries. A, et al, 2022).

A.I. for monitoring livestock wellbeing

Cognitive studies into affective states in livestock where being performed in 2021 as a way of mapping out animals’ emotional states. This can aid in improving animal welfare in the livestock industry in ways previously unavailable to farmers. Models have already been developed for use in human mental health and studies from this can area could be adapted for use in the livestock sector. This can allow for the development of more humane farming practises by looking at the animal’s emotional responses to treatment and how changes to this affects their emotional state (Neethirajan. S, 2021). Use of A.I. to monitor cow’s oestrus cycle has been increasingly essential due to a decrease in cow’s fertility since around 2001 which can cause significant financial loss for cattle farmers. Welfare measurement technology has become economically important now super markets pay premium for products which can demonstrate and validate good welfare practises. This includes proving cows spend an adequate time grazing as ill cows will spend less time grazing on average, in addition to this rumination patterns can be monitored so show signs of certain illnesses. Heat stress is another important area in welfare that sensors can monitor body heat compared to pen temperature as well as breathing rates. Precision feeding is a relatively new area in agriculture, feed mixer wagons can deliver feed mix data to cloud systems to record nutritional history and create more accurate diets for individual animals to aid growth rates, budget and welfare (Michie. C, et al, 2022).

Integration of A.I. and drone technology could allow for more efficient ongoing stock surveillance. Machine learning could also be used to detect and predict disease outbreaks for both crops and livestock. This kind of tech has been in use since around 2012 in cattle management by use of overhead cameras and vision algorithms. This allows farmers to view cattle in real time and alert farmers as soon as an issue arises. This can also include keeping track of births and signs of illness and strange behaviour as well as track the animal’s food and water availability. Automatic weeding could also be used to not only save farmers time but also keep livestock fields clear of potentially dangerous plants such as ragwort avoiding unnecessary fatalities (Sharma. S, et al, 2022).

Applications outside direct farming

One area of A.I. application is the development of “smart buildings” meaning buildings fitted with sensors to take actionable data and help modify systems within the building to better suits its occupant’s needs. One area smart building can assist in livestock is to help deal with fire damage which is a serious issue in farming as it can result in loss of animal and plant life on farms. It can also cause material weakening which even if the fire is put out can cause structures to collapse creating further dangers. A.I. can assist in this through linking algorithms to different types of sensors over wide ranges including vision-based detection and warning systems. IOT can also be integrated to assist in better fire prevention systems such as Realtime hazard detection and reporting. Several sensors can be implemented such as thermal imaging sensors linked to neural networks and deep learning programs. This can aid in evaluating explosivity and flammability of materials and substances on a farm and advice methods to minimise risks (Maraveas. C, et al, 2021).

Further development and potential setbacks

Machine learning could be used to track and predict weather patterns which could offset livestock storage and potentially impact livestock health i.e., hotter weather predicted more water should be allocated for animals. A.I. could be used to track ecological and environmental sustainability of nutrient cycles, this is key as a balanced nutrient cycle is key for sustainable livestock and crop production. A good example of this would be dairy farming, there are requirements for arable grasslands, sensors can be employed to keep a record of soil quality, grass yield and weather patterns. This can then be compiled with animal feeding levels and milk yields to see how they are affected by conditions the cows are kept in. A.I. could also be used to assist in breeding livestock such as in the case of gene manipulation to help improve animal traits this includes molecular marker assisted genomic selection. Digital hubs could also be set up to develop algorithms to measure and process data to characterise agricultural ecosystems allowing to see how micro-organisms, plants and animals interact within agriculture to create more efficient systems. One of the main focuses for future of livestock farming is to be able to produce more with the use of fewer resources and reduce waste. Precision livestock farming can be beneficial in this area, such as the sensor-based monitoring for quantitative data for life cycle analysis. This will make it easier to trace animal protein back to its source as well as data on sustainable crop production for feed (Holzinger. A, et al, 2023).

Application of agriculture 4.0 could lead to farming becoming more flexible in terms of production through monitoring real time interaction with people and products. One area at the moment that needs developing is the better understanding of agronomy and breeding processes among training A.I. programs for integration into farming (Amato. A, et al, 2022). In terms of data quantification agriculture is one of the hardest fields due to thousands of variables to consider. These include the great deal of data validation, collection and analysis for precision decision making for farmers (Linaza. M. T, et al, 2021). Contrary to popular belief it can be seen that further use of A.I. and automation in the agricultural sector can help provide further employment as shown in places like China and allows for solutions for emerging problems in agriculture. Agriculture is one of the most labour-intensive occupations and modern farmers are in danger of being overwhelmed while meeting rising demands. This alongside decreasing numbers of people going into farming due to perception of it not being a rewarding career path. Deep learning could help farmers predict trends in pricing patterns allowing for better budget in the future. It can also allow for them to identify sources of wasteful resource consumption patterns and help remove or adapt them (Javid. M, et al, 2023).

Conclusions

Looking at the materials covered in this essay it seems that A.I. could be highly beneficial to the livestock industry. Through smart monitoring systems to monitor livestock health can ensure high quality care for the animals and ensure that all farms are working to high levels of welfare standards and ensure livestock are provided an efficient diet and kept free of disease. Increasing automation on farms could help alleviate pressures on farmers placed on them by increasing global demands in an already labour-intensive industry. As such these measures could allow livestock farming to remain viable as the 21st century progresses by streamlining processes and aiding in better decision making in rapidly changing global conditions. However, most research at this time seems to cover the same points such as use of machine vision and large-scale data analysis with very little novel being looked into unlike other areas. This is most likely due to limitations on how A.I. can be implemented into livestock management coupled with the how recent efforts to integrate A.I. into livestock farming have been. Another drawback could be lack of universal integration of these technologies could be into the livestock industry. As it stands not all farms across the world use modern methodology as it is so may be unwilling to further modernise into farming 3.0. Many farms across the world still use traditional farming practises with some farms even existing for the purpose of keeping traditional practises alive and will be unlikely to adopt any new practises. Outside of these setbacks these technological developments will be a great boon to the livestock industry with the hope of further developments on the horizon.

Glossary

Deep learning: a method in A.I. which teaches computer networks to process multiple sources of data in a way similar to how a human brain would. This allows for the production of articulate and accurate predictions.

Internet of things (IOT): a collective network of connected devices and an online platform such as a cloud server which allows connected devices to share data to accomplish a shared purpose.

Machine learning: a way of training artificial intelligence which uses data streams and algorithms to train computers in a similar way a human can be trained to gradually improve accuracy of a specific task.

Nutrient cycle: a system where energy is transferred between living organisms and non-living parts of an ecosystem and then released back into the ecosystem through death and decomposition.

References

Amato. A, Amato. F, Angrisani. L, Barolli. L, Bonavolonta. F, Neglia. G, Tamburis. O, 2022. Artificial intelligence based early prediction techniques in agri-tech domain. Advances in Intelligent Networking and Collaborative Systems: The 13th International Conference on Intelligent Networking and Collaborative Systems (INCoS-2021) 13. Springer International Publishing.

De Vries. A, Bliznyuk. N, Pinedo. P, 2022. Invited review: examples and opportunities for artificial intelligence (AI) in dairy farms. Applied animal science volume 39. 14-22.

Fuentes. S, Gonzalez. C, Tongson. E, Dunshea. F. R, 2022. The livestock farming digital transformation: implementation of new and emerging technologies using artificial intelligence. Animal health research reviews volume 23. 59-71.

Holzinger. A, keiblinger. K, Holub. P, Zatloukal. K, Muller. H, 2023. A.I. for life: trends in artificial intelligence for biotechnology. New biotechnology volume 74. 16-24.

Javid. M, Haleem. A, Haleem Khan. I, Suman. R, 2023. Understanding the potential applications of Artificial Intelligence in agriculture sector. Advanced agrochem volume 2. 15-30

Liakos. K. G, Busato. P, Moshou. D, Pearson. S, Bochtis. D, 2018. Machine learning in agriculture: a review. Sensors volume 18. 2674.

Linaza. M. T, et al, 2021. Data-driven artificial intelligence applications for sustainable precision agriculture. Agronomy volume 11. 1227.

Maraveas. C, Loukatos. D, Bartazanas. T, Arvanitis. K. G, 2021. Applications of artificial intelligence in fire safety of agricultural structures. Applied sciences volume 11. 7716.

Micle. D. E, Deiac. F, Olar. A, Drenta. R. F, Florean. C, Coman. I. G, Arion. F. H, 2021. Research on innovative business plan. Smart cattle farming using artificial intelligent robotic process automation. Agriculture volume 11 (430).

Michie. C, Andonovic. I, Tachtatzis. C, Davison. C, Hamilton. A, 2022. Herdsman+: artificial intelligence enabled systems and services for livestock farming. Practical precision farming.

Neethirajan. S, 2021. The use of Artificial Intelligence in assessing affective states in livestock. Frontiers in veterinary science volume 8.

Patel. H, Samad. A, Hamza. A, Harahap. M. K, 2022. Role of artificial intelligence in livestock and poultry farming. Sinkron: jurnal dan penelitian teknik informatika volume 7 (4).

Sharma. S, Verma. K, Hardaha. P, 2022. Implementation of Artificial intelligence in agriculture. Journal of computational and cognitive engineering volume 2 (2). 155-162.

Wang. S, Jiang. H, Qiao. Y, Jiang. S, Lin. H, Sun. Q, 2022. The research progress of vision-based artificial intelligence in smart pig farming. Sensors volume 22 (6541).