The Ultimate Cobot

The Swiss Manufacturing Survey 2017 conducted by the Institute of Technology Management at the University of St. Gallen contains dismal findings: 46% of the questioned companies are considering a relocation of activities from Switzerland to a foreign country in the next three years.1 This, of course, does not include the businesses that have already done so. Their main destinations are China and Eastern Europe. The consequences for the Swiss labour market could be devastating. Yet, an efficient and flexible production through (partial) automation can reduce costs today to the extent that even low-waged countries cannot compete.2 Expensive and qualified workforce can take care of demanding tasks like high-skill manual assembly as well as the operation and servicing of equipment, while repetitive, low-skill activities are automated. Many Swiss companies still rely on manual assembly and logistic processes: production lots are too small, too diverse and some tasks are too complex for full automation. A human being is needed in the manufacturing process. Collaborative robots, or cobots for short, can safely work with human beings, showing great promise to increase the productivity of manual labour by exploiting the complementarity of human beings and machines.

• Great adaptability
• Capable of manipulating complex parts
• Complex and variable tasks
• Reliable execution
• Critical spirit
• Flexible

• Integral process control
• No ergonomic problems
• Can lift heavy loads
• Precise work
• Reliable execution of repetitive tasks
• Continuous work

A human being and a cobot may interact at three different levels [Figure 1]. Today, most installations rely on a separation of the workspaces: one for the human being, another one for the robot. This is the simplest way to satisfy current safety regulations. Still, this separation is much more affordable than typical robot automation cages, as it can be realized with sensors alone. This level is called human-robot coexistence. If the space is shared but the work is performed sequentially by either a human being or a robot, then it is called human-robot cooperation. Human-robot collaboration arises when the work is done simultaneously on the same task.

From the economic point of view, collaborative robots tend to be affordable (10 000 to 80 000 Swiss francs). Safety requirements are usually substantially lower than for conventional industrial robots, resulting in greater savings in nonessential equipment. Most importantly, reprogramming a cobot is simpler, hence no outsourced expert programmers are required.

The production may be laid out more flexibly. Cobots are capable of multitasking and appropriate for small to medium lots. They may act as third and fourth human arms while maintaining the expected around-the-clock availability.

Working conditions for human beings improve by relieving them from annoying repetitive chores. The task becomes more rewarding, improving job appeal. Human-induced errors are reduced by the introduction of a programmed machine.

Cobots may be installed to load, unload and tend existing machines, which have been designed to be operated by human beings. Functioning machinery and its fixtures can be retained, further justifying the investment.

Commercially available collaborative robots are easier to reprogramme than their industrial predecessors. Many of them allow the user to simply pull the arm to the desired position. This approach is only suitable up to human-robot cooperation as it does not allow implementing collaborative strategies.

We at the HuCE-roboticsLab are working to create interfaces to enable intuitive programming of cobots by the line worker. We are collaborating with several industrial partners that have defined use cases based on their production needs.

Typical challenges include the fact that a human being is now part of the process. Attention needs to be put on safety and the anticipated randomness of human actions. Collaboration may only arise if the cobot can detect the worker’s intentions and adapt its behaviour accordingly. Traditional approaches based on hardcoded trajectories are not suitable. New techniques based on online trajectory generation from contextual information are required.

Our research axes include recognizing context (situational awareness), identifying human intentions, allowing the human being to intuitively demonstrate and teach a task to a cobot and encoding the learned tasks for dynamic execution.

These developments shall lead to a higher manufacturing productivity and contribute to allow companies to maintain production in Switzerland, securing jobs and know-how for a competitive Swiss Made stance in the global market.

Different levels of automation can be defined depending on the number of identical units to be produced and their individual cost. Manual assembly is best suited for a low number of parts of high value. Fixed automation is the right choice for a very large number of parts of low value. There is a large span between manual and fully automated assembly. Two intermediate regions are identified one using Robots and another for Human-Robot Collaboration (HRC) [3]. Robot Automation introduces a certain degree of flexibility, as robots can be reprogrammed between longer periods of producing relatively large amounts of the same part. Collaborative Robots (Cobots) enable Human-Robot Collaboration with flexible production of smaller number of units for lower cost than pure manual production.

References

1 “Swiss Manufacturing Survey”, Institute of Technology Management, University of St. Gallen, 2017

2 “Erfolgreich Automatisieren im Werkzeugbau”, Studie Fraunhofer IPT, 2017

3 Matthias, B. et al. “Die Zukunft der Mensch-Roboter Kollaboration in der industriellen Montage”, Internationales Forum Mechatronik, 2013.