Author: Dr Conor Shanahan, postdoctoral research fellow, School of Biosystems Engineering, University College Dublin Staff at the Smart Systems Unit (SSU) within the School of Biosystems Engineering at University College Dublin, have developed a wireless sensor suite, Bosca, to monitor the internal environmental conditions of chicken houses. This research was funded under the Science Foundation Ireland Technology Innovation Development Award (TIDA) programme, in collaboration with Manor Farm, Ireland’s largest poultry processor. The prototype Bosca units were delivered in collaboration with Shimmer Technology. The UCD research team, headed by Prof Shane Ward, has developed and tested the Bosca units in commercial chicken houses within Manor Farm. The SSU specialises in the application of ‘smart systems’ within the agri-food and bioresource industries. Smart systems encompass the use of the latest technologies and systems associated with realising the potential of the ‘Internet of Things’, coupled with the widespread adoption of smartphones and the suite of capabilities delivered by them. Ireland is one of the highest consumers of poultry meat in the EU. Approximately 70 million chickens are produced annually in Ireland. Chicken producers in Ireland operate on very tight margins; hence it is essential that they optimise the performance of their production systems. Chickens are excellent feed converters but are very sensitive animals, easily affected by changes in environmental conditions. The in-house environment has major impacts on the performance of chickens, hence it is imperative that their environment is monitored and controlled to ensure chicken welfare, health and optimum weight gain. Chicken houses are very large, with a typical building having dimensions of the order 100m x 20m, and with some 20,000 or more chickens. There can be huge spatial variability in environmental conditions across the house, on a three-dimensional basis. Houses have environment control systems for heating and ventilation, but these gather data from only a small number of sensors, generally located at greater than 1m above the chickens. In effect, the system does not sense the environment in which the chickens live – i.e. close to floor level. The environment inside a chicken house is a dynamic one: chickens require higher temperatures when they are younger; humidity also has to be controlled in order to provide comfortable conditions to ensure the maximum growth rate. Carbon dioxide and ammonia levels can easily increase as the birds grow and these need to be regulated to prevent disease and foot ailments becoming a serious problem.

Bosca integrated 'smart' environment sensor unit

[caption id="attachment_17440" align="aligncenter" width="336"]Bosca A Bosca Mór (in green circle) hangs just above the chickens, monitoring the environment in the vicinity of the flock. An array of Boscaí Beaga units distributed around the house transmits environment data to the 'mother hen' Bosca Mór. These data are transmitted to cloud where data analyses are carried out to enable detailed interpretation of the data[/caption] All this points towards the need for a system that can monitor the environmental conditions within the chicken house, providing real-time spatial and temporal data from the chickens’ airspace, which is close to the floor. Standard systems do not do this, hence the SSU has developed Bosca, comprising a Bosca Mór, each with a number of Boscaí Beaga feeding data into the ‘mother hen’ Bosca Mór. The Bosca system is highly portable, being battery powered (a mains plug-in version is also available) and encased in a bio-secure casing that complies with cleaning protocols. It is an integrated ‘smart’ environment sensor unit and data management system. What is different about Bosca versus traditional environment monitoring systems is:
  • Its portability;
  • The range of sensors deployed;
  •  The ability to have different configurations of the sensors by the placement of Boscaí Beaga throughout the chicken house and in the chickens’ immediate environment;
  • Cloud-based information repertoires;
  • Web-based dashboard for data display;
  • Data analytics and the ability to add more modules to the Bosca unit.
The real value added arising from Bosca is the downstream data analytics. This is a separate operation carried out on the data gleaned from Bosca, and enables one to interpret the data – thus delivering enhanced understanding of the dynamics of the chicken production system. The Bosca has a number of sensors, which measure temperature, relative humidity, light intensity, carbon dioxide, ammonia and air speed. These sensors and the controlling hardware are housed in the Bosca Mór and Boscaí Beaga units, which are placed at floor level in order to sense the conditions within which the chickens live. For each deployment of the Bosca system in a chicken house, a Bosca Mór and a number of Boscaí Beaga are installed. The Boscaí Beaga comprise a smaller sensor array, containing temperature, relative humidity, airspeed and light intensity sensors. These take readings at a set time-period and transmit the data wirelessly to the Bosca Mór. The idea behind the Boscaí Beaga is to determine the spatial variability throughout the house. As the Boscaí Beaga log UNIX time and have associated GPS co-ordinates, they build a spatial and temporal map of the conditions within the chicken house. As the Bosca Mór receives data from the Boscaí Beaga, it transmits these data, along with the data collected by its on-board sensors, via the 3G network in .csv format to a repository in the cloud where it is parsed and stored in a SQL database. Sitting over the SQL database, there is a web-based dashboard for the display of relevant data for the farmer/producer, which enables one to compare houses on a farm or display environmental data from any of the individual Bosca Mór or Boscaí Beaga units in the houses. This is delivered via a web interface or directly by means of an app to the farmer’s smartphone. The system can send alerts to the farmer if any the environmental readings deviate from their tolerance range, allowing remedial action to be taken in a timely manner.

Data analytics for chicken flocks

[caption id="attachment_17442" align="aligncenter" width="448"]New Picture Drs Conor Shanahan and Patrick Jackman, from the UCD School of Biosystems Engineering, winners of the inaugural UCD Agri-Food Sprint Programme. Sprint is a NOVA UCD initiative in collaboration with the UCD Earth Institute, which aims to encourage the development of commercial outputs arising from UCD agri-food research by engaging with UCD researchers at an early stage in the commercialisation process. Bosca was chosen as winner for its technical merit and commercial potential[/caption] Similar to other ‘smart systems’, the Bosca produces copious amounts of data and these can be used subsequently for data analytics. Artificial neural networks have been developed that can accurately predict the weight of chickens 72 hours into the future, based on environmental data provided from Bosca measurements. The UCD Biosystems Engineering team has initiated an additional collaborative programme with the UCD Centre for Data Analytics, Insight, focusing on data analytics from Bosca and other data streams, such as ChirpMetrics (acoustics – chirp monitoring) and BirdEYE (visual and infrared movement monitoring). Additional modules for the Bosca are currently being researched through funding by Science Foundation Ireland. ChirpMetrics uses the vocalisations made by the chickens to determine the current state of the flock. This is achieved by recording the chickens and extracting an acoustic feature set from the vocalisations and then employing machine-learning algorithms, such as support vector machine and stochastic gradient descent, to classify the sound. This has already been tested and can accurately identify when the flock is hungry, stressed or happy. BirdEye uses thermal video images of the chickens in order to observe their positions relative to one another and key objects to identify general or local problems, including determining if feeding stations are being used efficiently and to identify potential gait problems.