Is the AI ​​market bigger than the robotics market?

Why is the robot revolution taking longer than expected?

Many analysts have predicted a bright future for robotics. Some even predicted that the number of robots installed will increase by a factor of twenty by 2025. But can you believe these numbers?

Industrial robots with grippers: Expensive, universally applicable types of robots with complex mechanics are predicted to show less growth in the robot market.

Photo: Jarmoluk / Bestgroup

While the industry is booming, the current numbers do not confirm the predicted significant increase. The landscape of robot companies has changed significantly in recent years. Instead of engineering-driven manufacturing companies for expensive, universally applicable industrial robots with complex mechanics and large payloads, the robot market is characterized by low-cost robots for a variety of applications or special robots. In Asia in particular, many new companies have entered the market. China has identified robotics as one of the main drivers of the Made in China 2025 strategy and will already have more than 800 robot companies in 2020.

How are the different areas of application developing?

New products are now coming to market quickly as there are many reliable component suppliers. Collaborative robots are on the rise. Open source robot operating systems and advances in artificial intelligence are driving additional growth.

On the other hand, when considering the application of robots in industrial settings, there are only limited areas in which the “mechanical ghosts” are ubiquitous. Body welding, gluing and assembly are key areas for robotic applications in Europe, mobile phone production in China and many service applications in Japan. In addition, the use of robots outside of industrial application has increased significantly. Medical and service robots were new key areas of application, while the number of vacuum, mopping and mowing robots has increased significantly.

A look at the numbers shows the success of the robot industry, but not the expected phenomenal growth. In analyzing the key challenges facing users, there are some obvious trends that will make it difficult for robot manufacturers to increase volume in the forecasted quantities. For greater success, there are some key innovations that can accelerate further growth in the robotics industry. This also gives rise to some important information for investors in this industry.

Trends in the number of installed robots per year in the most important regions of the world (data: IFR). Graphics: Bestgroup

Judge robots by the numbers

There are many very optimistic players in the market. One robot company alone predicts that it will sell one million collaborative robots per year by 2025. This would increase the volume of robots installed by a factor of three without a competitor selling a single robot. Assuming that this company would realistically achieve a maximum of 25 percent of the total market share, the total volume would increase tenfold.

According to the 2020 study by the International Federation of Robotics, the number of installed units increased by 12 percent in 2019, while the average annual growth since 2014 has been around 13 percent. Since 2010, the number of robots in Asia and Asia has increased by a factor of 3.5; in Europe only by a factor of 2.3, with America in the middle.

While a twenty fold increase would require a more significant increase, the growth of robotic installations has actually slowed. Looking at the annual figures, the number of robots installed decreased by 12 percent from 2018 to 2019 to 373,000 units. These figures do not correlate with overall economic growth, because, for example, Chinese industrial growth rose by 5.7 percent to 6.1 percent in 2019 compared to the previous year, while the number of installations in China decreased significantly.

Although enormous technical efforts have been made to bring robot costs down to a fraction of the original sales price, the expected increase in volume is lacking in many forecasts. Many newcomers have challenged existing robot manufacturers and some have grown significantly in recent years. Although the total volume has increased, many companies have still clearly missed their volume targets. This leads to many questions: Why are they lacking volume forecasts? Isn't the market ready? Or are the products too immature? What is hindering the adaptation of robots?

Why aren't robot sales soaring as promised?

In recent years, the robotics market has been divided into several key markets:

  • Industrial robots
  • Medical robots
  • Service robot
  • Consumer robots.

This list can be expanded in reverse order as follows. Consumer devices such as robotic vacuum cleaners, mops, and lawnmowers have seen significant sales increases in recent years and have grown steadily. Private end customers introduced one-task solutions very quickly and can now choose from a large number of providers. Low price, acceptable performance, and easy implementation have resulted in high levels of customer acceptance.

Service robots are also gaining more and more popularity. The original French robot called "Pepper" has been a well communicated model of robot for many years and is now sold by Japanese company Softbank Robotics. Different types of service robots are appearing. The Boston Dynamics robot dog "Spot" and the human systems "Atlas" and "Handle" provide an insight into the possibilities of the technology, as demonstrated by many YouTube videos. Service robots are being developed all over the world: Omron in Japan and Halodi in Norway are showing new concepts for the use of this branch of robots.

Easy-to-use and inexpensive service robots are on the rise.

Photo: Knight / Bestgroup

The volume of medical robots has increased significantly and competition is increasing. While Intuitive Surgical is the technology leader with its robots and covers a high price range, many competitors are starting to gain a foothold in the market. This includes many newcomers such as Johnson & Johnson, who combine Verb Surgical and Auris into one company, Medrobotics, Neocis, avatera medical, CMR Surgical, Corindus Vascular Robotics and others. Medical robots are facing a rapidly growing market. When exoskeletons like Rewalk Robotics, Sarcos or Ekso are added to this market, the result is a swarm of companies that are funded by large amounts of venture capital or large medical companies. As many countries have well-functioning healthcare systems, growth in this market is easy to predict as the systems significantly improve the quality of operations.

Robots in industry continue to establish themselves

Industrial applications were the starting point for robots. Seeing an auto production line with lots of welding robots is a widely accepted image. The application of glue, assembly and similar tasks have been carried out with robots for many years. In the past few years, new robots have changed the landscape. Universal Robots has created a new market with its inexpensive robots with limited payload but greatly reduced costs. Many competitors around the world have developed other low-cost robots, such as Franka Emika, ABB (in some areas), Aubo, Estun Robotics, Shanghai Step Robotics, Siasun Robotics and Flexiv. China has identified robotics as one of the key strategic drivers for economic growth.

Many new providers are currently entering the market in the field of low-cost industrial robots.

Photo: Becker / Bestgroup

In the industrial robot market, strategic alliances, e.g. the failed alliance Franka and Voith, and acquisitions, e.g. the purchase of Rethink Robotics by Hahn and now by Siemens, begin to redefine the market. A wave of concentration, as in other industries, is likely to be expected only when it is more ready for the market.

A special business area that has emerged in recent years is collaborative robots, robots that can interact with people and do not require the same safety equipment as other industrial robots. These robots have made it much easier to integrate robots into existing workplaces and open up many new areas of application. Industrial robots are now cheaper than a worker's annual wage and are actually an attractive investment. However, the actual number of installations does not meet expectations.

What is needed to create more salable robotic systems?

Although robots were originally created to replace human workers, the dexterity of human action is still superior to the most advanced mechanical journeymen. A person with all available sensors - feeling, eyes, ears - is far better equipped for many tasks, and the degree of freedom in one hand is far beyond the gripping capabilities of a robot. In addition, humans have the brain as a “superprocessor” that can deal with unforeseen conditions. Therefore, changes in technology are required in various robot components to increase utility.

A key area that requires significant change is the need for better sensors. Optical sensors that process stereoscopic views and enable three-dimensional evaluations offer a better control mechanism than the existing sensors. Because these only evaluate the positions on the many joints of a complex mechanical structure in order to monitor the movement of a complex structure. Control architectures that are fed with better optical input information enable better motion control of a robot.

At the same time, tactile sensors will gain in importance in the future. To date, there is little tactile intelligence in most robots that can enable completely new gripping abilities. The existing actuators are not well designed for robots. Robots ideally need a small motor with low speed, high torque, and low latency. Electric motors are larger than required and offer high torque at higher speeds. Applications in robots therefore require gears and additional brakes for fast reactions, which leads to large joint structures. Large joints lead to larger robots, while many applications require small robots.

At the same time, the structure of typical robots is more comparable to a human arm than to the two-armed human solution. While various structures - linear X / Y / Z systems, Scara systems, or hexapods - have been used to overcome some of the shortcomings of the typical six separate swivel joint systems, few robots have implemented two arm systems.

Programmable motion controllers have become smaller so that they can be integrated into robots. Therefore, fewer large control cabinets are required outside the robot. While the ability to control has increased significantly in recent years and artificial intelligence will expand the capabilities even further, not all tasks can be controlled well enough. Path optimization is more complex than the usual point-to-point movement, which is even more difficult to correct due to distributed intelligence on individual axes.

The next level of control - controlling multiple collaborative robots working on the same task - has not really been implemented as standard. This coordination is lacking even for two arms imitating a human being. While increases in volume will lead to economies of scale, the systems are usually quite expensive to manufacture due to the required size and precision of the mechanical structures.

With the “Robot Operating System”, an open source-based robot programming environment, robot manufacturers have a simple, real-time-based programming environment at their disposal. This eliminates many one-off costs when developing a complex control of the robot functions in the device.

Robotics will remain an area with a high rate of technical change, as many different segments will offer numerous innovation opportunities for “strokes of genius”. As previously described, sensors, actuators, control units and mechanical structures have several optimization requirements. Innovations are likely to lead to new system designs.

Changes to robots needed to advance their use. Graphics: Bestgroup

What is changing in robot programming?

Robot programming has changed significantly in recent years. While early systems were based on simple numerical control programming, this area has developed significantly in recent years. However, there are different programming languages ​​from robot manufacturers. A robot change requires learning a new programming dialect, in the worst case a new language. This means that companies have to make high investments in personnel for optimal robot care, but also for initial programming.

Programming itself presents several challenges. First a process has to be described. This can be done in different programming languages ​​with different support. Second, the process is multi-position oriented. Due to the inaccuracies in the physical robot structure, the key positions of the robot have to be taught to the system in real life, i.e. individually taught. This task will only be omitted when stereo vision with semantic understanding is implemented in the robots.

New programming environments offer new paradigms. In a “show-and-tell” system, a user moves a robot through all the steps, teaching process and position at the same time. With special motion libraries, standard tasks such as palletizing can be implemented much faster.

Google took a completely different route. With the “Everyday Robot”, the company uses its expertise in the field of artificial intelligence. The system can learn its own processes and learn how to program simple processes. While the system takes a long time to learn simple processes on its own, it can become very productive as soon as the learning from a robot can be transferred to all “siblings” and they know the same things.

Given the state of the art, robot programming deserves a "satisfactory" for applicability for most companies. Implementing robots still requires a lot of programming and is based on limited sensor input. Even a digital twin is not the solution for all needs, as the final system will require placement in positions in the final application if technology does not advance rapidly.

Implementing robots is still a challenging task

One of the key factors in using a robot is the ease of implementation. While robots were originally developed as variable, programmable products, many new robots are single-purpose applications with limited programming capabilities. And these new robots are selling better than their general-purpose cousins.

Why are the numbers of these service robots, medical robots and consumer robots increasing so quickly? Because they are easy to install. They do not require complex implementation planning and high additional investments. And their determination of benefits is usually very simple: These robots solve a task particularly well and offer a very simple profitability calculation. Each job of the robot saves a certain amount of money and over the life of the robot, the savings result in a high amortization of the initial investment.

Buying such a robot is a simple process: as soon as the robot is purchased and delivered, the amortization can begin quickly, as the implementation time is very short. The system may need to be adjusted once with little effort, but is useful from day one.

Industrial robots have a different challenge. One of the reasons for the low acceptance rate is the high effort involved in planning, implementing and commissioning the systems. The advantage of the universal solution is precisely the reason for this effort, namely the consequence of the complexity. Finding the best way to use a system takes much more effort than the single-purpose solutions described above.

For a typical industrial robot application, there is also a considerable need for a large number of peripherals. Safety devices, protective fences, material handling considerations, part feeders, special grippers and programming are some of the key elements required to achieve economical application and a long-term return on investment for the robots.While in the past the division between robot and integration was around 50/50 (in percent) for cost-intensive robot systems, the current low-cost systems lead to a ratio of 20 percent to 80 percent between the device costs and the total implementation effort. The cost of implementing a robot in a production environment is five times greater than the cost of the robot alone.

In order for multipurpose industrial robots to achieve a massive increase in volume, the installation problems must be resolved. Standardized part and standard interface designs, standardized containers, standardized gripping functions and maintaining the alignment of parts in robot cells are key elements in simplifying the planning of the use of robots.

Impact of the robot industry on production in Germany

The low-cost robot systems have a significant influence on the production strategies of manufacturing companies in Germany. They allow productions to be brought back from low-wage countries. Instead of transporting material over long distances to low-wage locations in order to benefit from the low cost level there, robot production offers an economical alternative. In this way, low-cost robot lines can achieve the same cost structure in Germany as some low-wage production.

In the context of projects for the assessment of delivery risks, more and more component manufacturing and assembly assemblies come to light, which have been relocated abroad for cost reasons with high logistical effort. With the new robot systems, some of these productions can be automated and then brought back to Germany. Together with the flexibility of the robots, on the one hand the variety of variants can be increased and, on the other hand, delivery times can be shortened. In this way, inexpensive robots will be able to further increase added value in Germany. Because in Germany there is a lot of engineering capacity for planning robot production lines. In doing so, however, the old optimization thinking of relocating to low-wage countries must be abandoned.

The increasing performance of robots and consistent line planning are creating new production and assembly concepts in which considerable potential for rationalization can be achieved in lines. Low investment costs for the robots in particular lead to short amortization times. There are still many challenges to be solved in this area, as robots are often planned as single-purpose production systems. How can robots be used to manufacture different generations of products? - This is an important task in order to strengthen the long-term use of robots in production and to make better use of the high flexibility of the systems.


Robotics will continue to see a “bright future”. Furthermore, many robot companies will emerge that will be successful in this huge market. The number of single-purpose robots will increase more rapidly in the next few years than in the case of multi-purpose robot installations. Currently, the time and effort required to plan and install robots will be the limiting factor in future growth for the universal systems. In contrast to the easy-to-use single-purpose robots, the complexity of planning, installation and commissioning prevents the use of universal robot systems.

While there are many opportunities to improve robot technology, industrial robot installations are not going to grow exponentially. Although the volume increase effects in the production of robots are very much appreciated by venture capital firms, the implementation effort at the user companies limits the scaling success. Without a significant improvement in the planning, installation and commissioning processes, not much will change in this area. If the average planning time for a robot drops from eight months to two months, a planner can implement four times more robots, which significantly increases the production volume of the robot.

Focusing on improving these processes will identify new approaches and require a much more modular system for aligning and handling parts. Feedback from such modules can improve the design of manufactured products to make them more suitable for robot assembly. This further increases the need for robots.

Even if the most aggressive growth forecasts are not achieved, the robot market will remain attractive for many years to come. Because even with a growth of 13 percent per year, attractive business models are possible for companies.

You might also be interested in:

Programming robots finally made easy - thanks to a new type of tool

Tips for safe human-robot collaboration

Dr. Torsten Becker works for BESTgroup Consulting GmbH and supports companies in the areas of innovation management, supply chain management and business development in new markets.