When is a robot really safe?

If the training of the employee and the complexity of the safety technology are in a reasonable relationship with the respective application, it is possible to speak of a safe robot. There is no such thing as 100% safety, but a very high level can be achieved if the right worker is assigned to an application and the safety technology is appropriate. Even the best safety technology is useless if it is manipulated in order to rectify errors or increase the cycle time of the system.

Where does safety start with robotics - with the robot itself or in the application?

Thomas Suchanek is responsible for safety and technical documentation at Yaskawa. © Yaskawa

According to the current state of the art, the safety of an industrial robot begins with the control system. I find it difficult to imagine an inherently safe manipulator in industry. It will always be a mixture of manipulator, robot controller and the appropriate plant safety technology.

Which safety standards apply then? After all, these are different for the control unit and manipulator.

For the manipulator as such, DIN EN ISO 10218 Part 1 is decisive for both the manufacturer and the distributor, TS/ISO 15066 also applies for collaborative robots and DIN EN ISO 10218 Part 2 for the safety design of robot systems, among other things. For the robot controller, there is DIN EN ISO 13849, Part 1, among many others.

So it's not the robot manufacturer who has to comply with the standards, but the distributor?

If a manufacturer is based within Europe, he must at least ensure that the product is state of the art and conforms to standards, unless - and this is the only exception - the manufacturer is not also the person placing the product on the market, because in Europe the latter is responsible before the law. This situation is defined in the Product Liability Act and not, as is sometimes mistakenly assumed, in the Machinery Directive. The latter defines the requirements for the safety of machinery, among other things.

Robots are often installed in a system network at the user's premises. Is it then the plant manufacturer who has to comply with the safety standards? How is liability regulated here?

According to Annex IV of the Machinery Directive, the robot is not classified as particularly dangerous machinery. This means that no third-party company, TÜV or accredited safety office has to approve the system. With appropriate proof, for example in accordance with DIN EN ISO 9001, a system manufacturer can independently place the system on the market and issue a declaration of conformity and a CE mark for the entire system.

What safety terms should everyone who is considering buying a robot know?

The keywords here are the Machinery Directive, the Declaration of Conformity and the list of residual risks derived from the manufacturer's risk assessment. Not many people are familiar with the latter in particular. No manufacturer is obliged to issue a risk assessment of a machine unless there is a justified request from an authority. If a risk assessment is done well, it contains many design aspects and (safety) interpretations, but no residual risks. However, the list of residual risks is essential for every operator. For example: If a robotic welding system is sensibly designed and also equipped with appropriate safety technology, then the list of residual hazards usually refers to any heavy, sharp-edged or hot workpieces. This means that a worker must wear appropriate protective clothing for removal.

As a rule, if you make a machine as safe as specified, you can no longer use it or its performance is severely restricted. A classic example is the hand-held circular saw: here, the saw blade is protected by a guard that opens as far as the penetration depth of the saw blade into the material in question allows. This minimizes the danger posed by the saw blade. If I wanted to meet the requirement of "no dangerous movement when the operator is present", I would have to enclose the entire saw blade. But then I would no longer be able to cut with it.

This is the crux of the matter that every machine and plant manufacturer has to deal with: How small do I have to make the safety window in order to guarantee the function of the machine and still not violate the safety concept?

When does a robot become a safety-critical system?

If the safety-related design and characteristics promote a process that is too complex. In the case of a robot system, this means that a robot works in a delimited area behind a work table and the worker has to insert and remove the components from the front or transport them away at the rear of the processing station. If the safety technology goes beyond this, i.e. if the suspenders and the safety staple are also used in addition to the belt, the safety technology can lead to manipulation attempts. In this case, this would be an employee attempting to circumvent the safety equipment (protective housing) by disregarding the relevant steps such as acknowledgement and release routines in order to restart the machine in the event of a fault. This creates a hazardous situation that could have been avoided with the right training and the right safety technology.

What is the awareness of safe robotics on the market?

In the last four to five years in particular, not exclusively due to, but at least driven by the market launch of collaborative robots, an understanding of certain safety requirements has become established. However, it is not yet on a par with the economic and production-related benefits that can be achieved by using a robot in production.

What safety requirements does Yaskawa face on a daily basis? What protection mechanisms are required?

These are the classic fixed guards, but also non-contact protective systems such as light barriers, laser scanners or access control from the ceiling. The market is demanding a reduced footprint of the system and increasingly fenceless production. Without perhaps having collaboration in mind, companies do not want fixed safety technology such as safety fences or sheet metal enclosures in order to gain a few more square meters of work and hall space.

What role does security play in terms of safety, also with regard to Industry 4.0 and networked production?

An aspect that has often been asked about! We have the advantage that our control systems are developed in Japan, which means that the protocols are less well known on the European market. This means that the security concept offered by an open robot controller is not yet as pronounced as it might be with European competitors. Security will play an important role in the future, but the criminal energy must be very high: first you have to hack into the system and have the corresponding knowledge of which position needs to be changed so that you can cause damage (materially and unfortunately also in terms of security) via certain barriers, passwords and checksums. I see more of a danger here of gaining very specific specialist knowledge quickly and therefore cheaply in order to gain more than unfair competitive advantages.

Everyone is talking about human-robot collaboration. How does it differ from human-robot collaboration and is this difference widely known?

The four defined types of human-robot collaboration. © Yaskawa

The difference is at least comprehensively defined in TS 15066. The industry is actually very interested in human-robot collaboration realized via a force torque limitation of the robot. An HRC 2, for example, is a human-robot collaboration in which worker and machine share a workstation, but never at the same time. I don't believe that the difference between cooperation and collaboration is fully understood, and it will be several years before the industry agrees on a truly fine-grained terminology. Many people who talk about human-robot collaboration actually mean cooperation and this does not necessarily have to be realized with a cobot.

Can an industrial robot be made so safe with software that it is suitable for HRC, and if so, how?

This is certainly conceivable, but you have to take a very close look at the geometries and forces exerted by this robot. The actual human-robot collaboration is based on the assumption that the pain threshold is not exceeded. In comprehensive studies, we have approached values in the standards committee and accordingly defined a range that is tolerable for both transient contact and quasi-static contact. There may well be manipulators which, due to their design and their very low load-bearing capacity, can only keep the contact so low that it is below the pain threshold when monitored by a safe control system. However, whether the robot can still be operated with cycle times acceptable to the customer is a completely different matter.

So a yes with restrictions?

No, a yes with precisely defined specifications.

What are the safety trends of the next few years?

It's like looking into a crystal ball. The industry is asking more and more for fence-free production, and robots need to be easy to operate. Due to the shortage of skilled workers, more and more unskilled workers are entering production so that the required quantities can still be guaranteed. As a result, the robot itself must have a certain level of intelligence to compensate for the lack of knowledge and experience of unskilled workers. In this context, safety trends will be based on the level of training of humans. Traditional safety components will increasingly take a back seat and new technologies, methods and materials will be used. These will also have to comply with the requirements and specifications - at the level of the "state of the art" and standardization.