Cobots work directly with people without a safety fence. They are lightweight, transportable, easy to program, adaptive, flexible to use and space-saving. The associated infrastructure is compact. However, these advantages bring challenges in terms of safety. Because:

How can flexible collaboration be designed safely? The prerequisite is a comprehensive, application-specific risk assessment. In the case of human-robot collaboration (HRC) without a safety fence, this is complex. It requires those responsible to have precise knowledge of the applicable standards and laws. Independent testing organizations can provide support if they have experience in the assessment of robot systems and knowledge of the planning and implementation of cobot applications. It should always be noted:

The design of cobots is fundamental to their safe use. The technical specification "ISO/TS 15066 - Robots and robotic devices - Collaborative robots" defines safety requirements for them and their working environment. It also supplements the normative requirements and instructions in ISO 10218-1 and ISO 10218-2 for the operation of collaborative industrial robots. For collaboration between humans and cobots, "EN ISO 12018 Industrial robot safety requirements" defines four basic safety objectives:

1. a safety-oriented, monitored standstill. 2. human control of the robot. 3. speed and distance monitoring to prevent dangerous human-robot contact. 4. power and force limitation if contact is required.

If the final safety check confirms that it is safe and legally compliant, the robot application receives a CE mark.

The monitored standstill and other safety objectives are implemented using electro-sensitive protective equipment (ESPE). As soon as the distance falls below the required minimum distance, the ESPE - light curtains, laser scanners or pressure-sensitive systems - react. According to "DIN EN IEC 62046 Safety of machinery - Application of protective devices for detecting the presence of persons", they must be tested during commissioning and annually. In addition, the systems must be inspected regularly for damage or soiling. The entire installation situation, the triggering process and the overtravel must always be included in the safety assessment.

Flexible safety zones, biomechanical limit values

BWS are not very adaptable. They always react in the same way when they are executed conventionally. If the distance between the employee and the cobot is less than the programmed minimum distance, an ESPE stops the machine and interrupts the work process - even if the human and machine are already moving away from each other. These illogical patterns can be reduced, for example by using the dynamic safety zone approach. Safety systems analyze and evaluate the behavior of humans and cobots and the resulting hazards. However, it is almost impossible to predict the interactions during operation, especially when environmental and process conditions are variable. In future, software-based approaches should overcome these challenges.

There are specific risks of injury where cobots and employees have to come into direct contact. Manufacturers and operators should take sensible countermeasures based on specific measurements of the forces that occur or combine various measures - such as reducing the speed, limiting the forces, increasing the contact surface or a partial enclosure. In addition to contact with the upper body and arms, hits to the head are particularly dangerous: The BG/BGIA recommendations for "Risk assessment in accordance with the Machinery Directive - Design of workplaces with collaborative robots" specify biomechanical limit values: 90 Newtons as the maximum permissible impact force, 20 Newtons per square centimetre as the maximum value for surface pressure.

Ensuring safety in line with requirements

The adaptive safety of cognitive production systems is a concept - developed in the context of Industry 4.0 - that is of particular importance for HRC. The basic idea is to ensure safety as required without restricting production as much as possible. Adaptive safety of cognitive production systems is implemented with the help of learning (multi-agent) systems with algorithms for controlling sensors and actuators. In order to ensure the safety of employees at all times, even when they are working closely together, the systems should predict the development of dynamic processes and control the cobots accordingly. Availability is also important in smart factories; production systems are assembled on a modular basis as required. The modified system configuration usually has to be reassessed afterwards, which is often associated with production downtimes. Automated certification processes can minimize these effects by determining whether the interlinked machine system complies with standards at runtime if possible. At the end, systems should automatically issue approvals for the entire system - according to clearly defined rules.