Human-robot collaboration: Symbiosis instead of contradiction
With the rapid pace of technological development, robots of all kinds are increasingly becoming a topic of conversation at the family table. Some think very little of robots and see them as somewhat of a threat to humans. Be it as a «superior» workforce or even as potential bringers of an apocalypse. Others see robots as an opportunity to combine the abilities of robots and humans, thus combining the best of two worlds for the common good of all. This is exactly one of the most central goals of Human-Robot Collaboration. So one could also summarize in short form that robots should ensure that humans no longer have to behave like robots. An noble objective, isn’t it?!
What is Human-Robot Collaboration?
The Human-Robot Collaboration, as the name suggests, is all about the cooperation between humans and robots. The robot is intended to support the human being in his work and in particular to relieve him of static, repetitive, strenuous and sometimes physical work. Examples include copying thousands of data values from an Excel worksheet to another IT system or lifting and moving heavy objects. More complex tasks requiring a high degree of precision are also conceivable, such as support in industrial manufacturing processes, for example. In industry, as you certainly know very well, robots have been playing a central role for quite some time already. In the automotive industry, for example, the first robots were used by General Motors as early as 1961. However, as technology has developed, so have their applications. Today robots play a leading role in the optimization and redesign of production processes.
Robots that work together with humans are often called «collaborative robots» or «Cobots». They work with people at the same workplace and therefore generally require less floor space. The separation of the workplaces of robot and human is thus removed. The Cobots are operated with sensors that react to human touch. The safety, which was previously provided by a fence, is now ensured by operating with these sensors and by reducing the force, speed and strength of the robots.
Apart from physical work, there are also other possible applications of robots, as already mentioned. In other words, there are now thousands of robots outside of production facilities too. Today, their working environments include more complex environments such as homes, offices and hospitals. But of course also computers and the Internet. As described in a previous article, basically all those tasks can be automated that are rule-based and occur repeatedly. A prime example of such an application is OCR, or Optical Character Recognition. More on this later.
Advantages of Human-Robot Collaboration
Human-Robot Collaboration has numerous advantages to offer. Certain advantages are specific and can only be achieved with the respective application. But there are also advantages that apply across all possible applications.
Human capabilities such as solving complex, cross-contextual problems, creativity (although we can argue here, since randomness can be wonderfully configured in a system – so it’s merely a question of definition) or more sophisticated communication are still not really automatable and thus not robustly executable by robots. It will probably take a while before a robot has such complex capabilities. In various disciplines, such as communication, however, we are making great progress – see OpenAI’s GPT2/3.
Therefore, the argument that robots are a major threat to humans as a workforce can, at least today, be strongly refuted. Of course, there are many jobs that meet the above-mentioned criteria, i.e. are repetitive and rule-based and can therefore be automated at least in part. But if we are really honest with ourselves, these are most likely not really good and attractive jobs either. And there’s plenty of work to be done still.
As a rule, employees are therefore not dismissed and are rarely ending up unemployed as a result. In my opinion, the human labour force will have a different structure rather than being completely replaced. Thanks to the robots, employees have less to do with their previous work and more time for other kind of work. This means that employees can take on new responsibilities during this time, which allows them to be more human, creative and inventive and to cultivate more social contact. Such work can ultimately lead to higher job satisfaction. In the spirit of the comparison mentioned at the beginning: robots should ensure that humans no longer have to work like robots.
One of the main advantages of Human-Robot Collaboration or Robotic Process Automation (RPA) is therefore the resulting increased job satisfaction and the potential that it offers, which can take on all kinds of forms and shapes. Whether robots at the workplace are a curse or a blessing is a matter of dispute even among scientists. If you want to go deeper, you can find more information here.
An additional complementary cross-industry advantage is clearly the increase in efficiency. Robots never get tired, never have to go to the toilet, never take a lunch break, never get sick or demotivated and never go on holiday. They are therefore undoubtedly more efficient workers than people. The increased efficiency thus also leads to relative cost savings.
In addition to the increase in efficiency, robots also have the advantage of making few to no mistakes (to remain reasonably compliant statistically: <0.000001%), whereas humans make mistakes much more often. Mistakes that happen early on then continue until someone notices the mistake. This can quickly become complex and expensive.
Human-Robot Collaboration in document extraction
Now as promised on the subject of OCR. OCR can recognize the characters on documents and then convert them into digital formats. For example, the data on invoices, delivery notes or contracts can be read into a central document management system (DMS). This capability is extremely valuable for companies that receive high volumes of documents, whether digital or physical, and want to store them in digital databases or process them in downstream processes without having to spend hours entering data manually. More about the topic (RPA) and OCR can be found in this blog article.
For example, an accounting expert or a fiduciary could upload an invoice to an OCR software interface and the technology could then recognize the characters on the document.
Thanks to machine learning, OCR technology can recognize a wide variety of different documents despite differences in layout. If you have ever worked with old-fashioned OCR software, you know what a game changer this is. For companies and their employees, this means less tedious, monotonous manual work and more social interaction and demanding work, which of course has significant economic effects.
However, the work cannot be completely outsourced to digital robots in this context. Definitely not yet. Because these technologies also have weak points. For example, extraction robots can make mistakes in the analysis of the data. If a part of the text has been marked with a highlighter, it is much more difficult to extract the text without errors. Similar challenges are presented by background images, watermarks or background colors. If the documents have not been scanned cleanly from the outset, this can also lead to difficulties in the extraction process. For example, it may happen that the letter “S” is extracted instead of the number “5”. Even if there are calculation errors on the document, these are only extracted but not corrected. So all it needs is a little noise and the job is much more difficult for robots.
Because of this, human post-processing can be extremely useful. Such mistakes can be corrected after the robot has done its job, thus preventing devastating consequences. This human-robot team, i.e. the training of the robot, will also help to minimize such errors in the future. This thanks to machine learning technologies.
This is exactly why we at Parashift are so keen to focus on the Human-Robot Collaboration. Cause state-of-the-art machine learning technologies combined with human post-processing enable a highly flexible and efficient solution that is up to any document extraction challenge, always delivers excellent data and, above all, is future-oriented.