KANBrief 3/22
The development and use of highly automated machinery is growing owing to the efforts of agricultural machinery manufacturers and users in the most diverse of sectors. The German Social insurance for agriculture, forestry and landscaping (SVLFG) plays an active role from an early stage in the shaping of new technologies, with the aim of influencing them in the interests of occupational safety and health. The key objective of prevention in this context is to protect persons against hazards presented by highly automated machines.
In the future, highly automated guided vehicles will be common in agriculture and throughout the green sector. At present, two essential work environments are distinguished in agriculture: the farmyard and the field.
The farmyard work environment includes, for example, automatic feed systems (AFSs), manure scrapers and feed pushers. The self-driving or automated guided vehicle moves around different parts of the farmyard (livestock shed, silo storage area, yard area). The risk assessment must take account of requirements for use both indoors and outdoors.
The associated technical components such as feed bins, conveyor belts, mixing bins, discharge systems, etc. are often interconnected. Before a highly automated guided vehicle is put into service, a declaration of conformity for the entire installation in the agricultural enterprise must be produced in accordance with the provisions of the Machinery Directive. Without this declaration, the vehicle may not be operated.
For the field work environment, particular consideration must be given in the risk assessment to the higher travel speeds of automated or self-driving agricultural vehicles. For this application environment, tractors are available with highly automated functions, and may or may not feature a driver’s seat. Further self-driving machinery without a driver’s seat and capable of highly automated guided travel also exists. The range of vehicle types is broad. Studies of the market reveal products ranging from large tractors with over 300 horsepower to minute robots performing highly automated field work.
Environment recognition is particularly important for any automated or self-driving vehicle. Decisions that were once the preserve of the farmer are now taken by the vehicle manufacturer. Detection of persons, objects and obstacles within the environment of the work to be performed must be ensured:
The combination of tractors and attachments is a major issue. Where an attachment may be fitted to the base vehicle and the attachment significantly exceeds the base vehicle’s width or turning radius, it is not sufficient for the manufacturer of the base vehicle to limit environment recognition to the direction of travel. This may result in collisions with persons in the direction of travel. Initiation of travel may also be a source of risks. Before the vehicle starts to move, it must be ensured that no persons are present either in the direction of travel or between the tractor and the attachment. This requires the environment of the base vehicle in combination with the attachment to be monitored, and not merely the environment of the base vehicle alone.
Sensor technology is of key importance in environment recognition. The SVLFG takes the view that certified systems should always be used for the detection of persons. Most modern object recognition systems are not suitable for ensuring that highly automated guided vehicles are operated safely. A distinction must also be drawn between person detection systems for indoor use and systems for the outdoor environment: the latter presents a much greater challenge. Changing light conditions, rain, snow, leaves and dust are among the many factors that the environment recognition system must reliably register and evaluate. This can often be achieved only by a combination of sensors.
Automated guided vehicles that are used in safeguarded areas of the enterprise that are not accessible to persons can be compared to automated manufacturing equipment. Measures must be in place to protect persons entering parts of the safeguarded area, for example for the purposes of troubleshooting or maintenance. Under such circumstances, vehicles and other automated parts of an installation must be placed in a safe idle state. Components may then be moved – if at all – only individually and at reduced speed by manual control (maintenance mode). Departure from the maintenance mode must be possible only by means of manual confirmation effected from outside the safeguarded area, and only once it has been vacated by the persons entering it and after the access points have been closed.
The SVLFG is involved in the current revision of EN ISO 18497:2018, Agricultural machinery and tractors – Safety of highly automated agricultural machines – Principles for design. The standard is to be restructured and developed further as follows:
The key aim of prevention is the reliable detection of persons. Protective equipment that deactivates the machine only when it comes into contact with a person (bumpers) is no longer acceptable on its own, especially for machinery with higher travel speeds. In view of the possibilities now offered by artificial intelligence (AI) and state-of-the-art environment recognition, the occupational safety and health sector now regards contact of such a machine with persons as unacceptable.
ISO 3991 governing the safety of robotic feed systems is currently being drafted. Prevention experts from the SVLFG are also involved in this project, in the interests of safety and health in the future.
Sebastian Dittmar, German Social insurance for agriculture, forestry and landscaping (SVLFG)
Sebastian.Dittmar@svlfg.de