This relates to productivity and optimizing the use of farm inputs, as well as to energy consumption. Increases in efficiency while retaining constant machine size are achieved by more intensive use of sensors and electronics:
◦expanded and new sensors capture changing growth conditions of crop plants more sensitively, precisely and efficiently;
◦high-tech machines and their sub-systems adapt automatically and quickly to changing conditions of use and operating modes.
Parallel with machine developments, ‘Precision Farming’ has provided fundamentally new methods in many areas and thus contributed to further progress in agriculture. When precision farming was introduced, the principles were not new, but it was only the availability of techniques using the Global Navigation Satellite System (GNSS) that made new methods technically feasible and partly fit for complete automation. Precision farming has made it possible to describe and analyze plant properties and their growth conditions better and even to predict events, for example in plant protection, and respond with selected measures. Outstanding innovations provided by precision farming included the ability to supply generally reliable information on the heterogeneous production process and make precise and variable application techniques available.
Structuring general data management on the farm simply and compatibly
However, what precision farming has not so far achieved is to make general data management on the farm compatible and simple. Space-related and time-related data together with other relevant information could not be integrated reliably into new strategies for crop or farm management. Linking distributed, various, public and farm information sources is evidently still problematic. These make high demands of information management. Parallel with this, there is a lack of automated decision-making aids that use networks of different information sources to support and optimize production processes. And it is precisely here that Smart Farming aims to tackle the problem and develop precision farming further. Smart Farming is based on precision farming and also relies on information-intensive technologies. However, it makes much greater use of context-based strategies at a higher level of knowledge and automation than precision farming has done so far. Context-based here means that information is never processed in isolation, but can always be linked and processed in automated fashion together with other information. The precision farming techniques in recent years relied too strongly on data from individual sensors and individual information sources. Combining and fusing various relevant sensor data is necessary today in order to optimize complex systems such as managing crops or transport logistics of harvested produce on the basis of specific criteria.
Need for research into data fusion
On the grounds of the complexity of the tasks involved there is a need for research with a view to linking information (data fusion) in order to tap hitherto unexploited potentials of precision farming. Data fusion must be automated, as other solutions would create too much extra workload for the farmer. For example, there is still potential for increasing yields with optimized fertilizing strategies. Potential solutions applied to date do not go far enough, because they often only act with single parameters. In concrete terms, this means that real time decisions should not be taken on the basis of information from just one sensor type. Particularly with a view to automation, the goal must be to capture key parameters in real time or via other sources and pass the information on for suitable data fusion. This means that having important and valuable information available does not help much by itself if the linkage algorithms are not known and have therefore not been defined.
The same applies for complex machinery controls that may not be left to relate to just one measuring quantity. What are required here are strategies that support the operator and offer him genuine options for choice in order to weight process steering in different ways and thus optimize it specifically. This covers the option of specifying objectives for machinery use, such as for example harvesting with "loss minimizing" or "time minimizing". Here the automation must be available to combine not only machinery data but also process data and external information (internet) and suggest options to the operator. Such improved operating concepts with adaptive capacities are interesting even for experienced machine operators of very complex machine systems in order to develop and exploit unused potentials.
Systems and their components for Smart Farming should display the following attributes:
◦Knowledge-based and context-based
Knowledge can only be generated if information is linked and set in a context. Algorithms (mathematical modelling) to describe interactions between sources and types of information are necessary for automation.
◦Interactive
Unrestricted communication between internal and external systems is the prerequisite for exchanging information.
◦Adaptive
Systems become adaptive thanks to self-regulation with modifiable strategy specifications of the operator.
◦Transparent
Systems do not represent a blackbox. The basic functions of the techniques must be explainable and understandable, otherwise they do not generate any confidence or any acceptance among operators.
◦Automated
Systems display a high level of self-reliance. If problems arise, however, the operator is informed and possibly requested to take decisions when the automatic system reaches its limit.
Examples of how solutions are described or how approaches to precision farming could be developed further:
◦Automated data collection
In Germany and the rest of Europe it is quite normal for tractors and mounted implements to come from different manufacturers. Electronic controls for mounted implements are increasingly being used and are available in standardized form via the "ISOBUS". The potential of this system is, however, by no means exhausted with machine controls. Using automatic data collection and transmission and storage of information from on-farm processes on farm computers is promising. However, if data do not correspond to the ISOBUS format, this makes use of the machinery and equipment more difficult. Data that may perhaps already be available on the farm computer – such as for instance digital soil maps – cannot be combined with other data without additional work input. Generally, it would also be expedient to integrate off-farm information, for instance from suppliers and customers, contractors and machinery dealers, as well as requirements made by statutory specifications governing actions and administrative regulations (observing laws and certification requirements).
◦Real-time sensors for crop management
Methods of combining mapped information and parameters measured in real time with the aid of mathematical models in order to consider and compensate the pros and cons of both data capture systems are very promising. These systems generally only cover biomass and chlorophyll density, although other also varying parameters within a field such as growth stage, crop density (sowing rate), shoot and ear densities, variety type, plant diseases and weed pressure would also be important for optimal crop management. Basically, a plant crop represents a very complex system that cannot be adequately described for the purpose of nutrient needs by collecting just one parameter. As long as the most important relevant parameters are not captured, whether recorded in real time or mapped building on other sources, automated fertilizer application cannot represent an optimal solution for a crop field. Alongside sensors for recording crop parameters, equipment for recording and describing soil physical and chemical properties should also be intensified further.
◦Spatial resolution of application techniques
Increases in capacity with wider working widths have led to small-scale structures within the working width no longer being workable. Especially when tramlines no longer run at right angles to the headland, excessive dosing or failure to apply agents on partial areas can occur. Here, precise positioning systems have rectified the situation by automatically switching boom sections on and off and thus avoiding incorrect doses. This can save sometimes up to 18 % in plant protection agents.
◦Coordinating machines with optical sensors
Optical sensors such as cameras can lead to essential relieving of the operator's workload with field choppers and harvest removal trailers. Optical identification and location of windrows can be used for automatic steering of forage choppers, while recognition of trailer dimensions and filling levels serves to control the position of the ejector manifold and flap. Camera systems on the exterior are difficult to realize due to changing light conditions and high computing performance required in real time. This system represents a new quality in machine controls. It helps to minimize harvest losses and relieve the workload significantly for operators of choppers and tractors.
◦Similar advantages can be noted for automatic headland management when the tractor automatically assumes complete control of the machine during turning, including control of mounted implements.
◦Automated steering systems with optimized track guidance
Automatic steering systems are a commercial success. As they minimize areas of overlap, they save fuel and time and relieve the burden on the driver. Moreover, with these systems it is possible to select completely free track guidance or have it optimized by a computer. This reduces turning times and routes. Studies have shown that these systems can save over 40 % time and diesel at the headland.
Today, decisions in plant cultivation are based essentially on space-related and time-specific information. Methods using single information systems are not well accepted and will be increasingly replaced in future by more automated knowledge-based systems that are distinctly easier to use. A variety of sensor techniques as well as all available on-farm information must be made usable for a system in order to allow individual, prompt and efficient decision-making aids. So far, techniques have not been able to fully exploit this potential. Smart Farming would be a great advantage for results-oriented and resource-conserving agricultural production. The operator or farmer with his exceptional qualification remains at the hub of all on-farm processes to make decisions. Through improved information techniques and automation, however, he has more time for what is essential, as routine work and methods fall by the wayside. He is kept constantly informed of processes and reliably supported by Smart Farming in his decisions, enabling him to optimize results altogether depending on his farm management strategy
