A Belarusian Company Built a Collaborative Robot
A Belarusian Company Built a Collaborative RobotIt took Rozum Robotics, a Belarusian startup, a year and a half to build collaborative robot, which is now available for pre-ordering.
The Belarusian invention costs about 20 000,00 USD (plus a 20% discount for the Belarusians). The exact price can vary, depending on the model, but it is still about the third of that offered by competitors. Nearly all components of the cobot (collaborative robot) are self-designed. What Viktar Khamianok—the founder of Rozum Robotics—used to ironically call “The Skilled Hands laboratory” has now turned into a manufacturing business with its own development hub.
By the middle of 2018, the company plans to release about 20 cobots, subsequently manufacturing 10–15 of them a month.
No rocket scientist needed to program the robot
In the office, the Rozum Robotics team demonstrated how the robot signed in to Google Chrome, opened the company’s website, and ordered itself by tapping the order button on the touch screen.
“Programming our robot is easy, even if you have no head for science,” comments Andrei Shaurko, as he drags the green command blocks—Add Point, Close Gripper—across the screen of a rugged tablet PC (the software is based on Blockly by Google, an open-source library for creating a visual programming environment). “People from competing companies envy our simplicity, complaining their solutions are often over-engineered. It’s true that their software often seems to be sophisticated enough to program a spaceship. Rozum Robotics makes its robots for those who’d rather not deal with complicated coding.”
If you set a small resolution (accurate to the tenth of a millimeter), the robot can assemble, for example, a servo motor for itself. In total, the robot includes six servo motors functioning similar to human joints.
At first, the software would fly out of control occasionally, and the robot would lose calibration (e.g., execute the same program, but in a different plane). Today, the software is stable, and the robot obeys its creators perfectly. You can control it from any gadget, including a smart watch.
“We were the first to come up with a business case for it,” jokes Andrei. “Suppose, you are an Operations Manager, and you have to attend a meeting on the 5th floor. With the watch, you don’t have to carry a tablet around. You’ll just get messages like: “Hey! I’ve just finished my task, give me some more!” or “I’ve hit an obstacle! Help!” ”
How to leave a 10-year-old business for a robot
Viktar Khamianok started programming in the 5th grade. By the 7th grade, he had read a whole library of books, varying widely in scientificity. So, it was only logical that he’d choose “Artificial Intelligence” over other specializations in the Belarusian State University of Informatics and Radioelectronics.
A spelling mistake, however, in a math test caused Viktar Khamianok to abandon the dream for “something more real.” The same spelling mistake brought him together with Oleg Borisevich, leading to establishment of BelPrime, a product IT company.
In fact, the initial idea was to sell outsource services, but it happened so that the company found no suitable tool for promoting its website, when it needed one. Therefore, the developers decided to build the tool on their own, to speed things up. Ten years later, one out of every third marketing specialist in the world used the BelPrime’s software.
“Maybe, I should blame the midlife crisis, but, one of these fine days, I just realized that Internet marketing software was not the thing I’d like to do for the rest of my life. For that matter, I’ve been dreaming about robots since I was a child,” recalls Viktar. “Now is just the right moment: robots will soon change our lives completely, just like computers did in the 1980-1990s. So, it makes sense to tack up the horse to lead the progress. Plus, it gets even more interesting because the results are tangible. Unlike software, you don’t just see, but you can also touch the hardware you built. As you breathe life into bare metal (e.g., installing firmware) and watch it coming alive, you can feel the power of creation in your hands. It is inspirational experience, indeed.”
Back then, Viktar had a vague idea of robotics—mostly based on his childhood fantasies and the “Terminator” films. At first, it was really just a “Skilled Hands laboratory,” as Viktar calls it ironically. In April 2016, Rozum Robotics finally appeared as a company.
Leonardo helped to sort things out
Leonardo, the startup’s first-born 3D-printed robot, is now getting dusty on the top shelf at the workshop in the Rozum Robotics office. For the startupers, it is already a museum showpiece.
Robots, such as Leonardo, are said to soon replace cartoonists in parks and at beach promenades. Fitted out with a camera, the company’s first robot can paint a person’s face from a photo or from life.
A three-year-old child has a better technique than Leonardo, but the whole thing was not about creating a masterpiece. Working on the complex task of getting the robot to paint, the team studied the mechanics of a robot and learnt how to control it.
“There are two sides to the robotics. One of them is amateur engineering, cheap solutions built from AliExpress components, and the other—engineering of industrial robots,” comments Viktar. “Leonardo is an amateur project. Its servo motors cost 350 Euros each, and it has 10 of them in all. At first, it seemed really expensive. However, it turned out that all you could build for the money was a matches-and-acorns handicraft to amuse your Mom. For a robust robot, you need a motor that costs around 5,000 Euros. Our robotic arm has 6 of them, which amounts to 24,000 Euros—the cost of a brand-new car. It is too expensive for robots to invade the industrial and household markets at large scale.”
Why self-designed components, not Chinese
Initially, Rozum Robotics had no intention to build servos or electrical motors or to make other, more sophisticated components, such as chips, with all the maths and complex algorithms involved.
“We thought we’d buy the components from reliable manufacturers, assemble them, develop our own software, wrap it all up beautifully, and put it up for sale. Why bother building electrical motors when you can buy a Chinese one, which are so numerous?” recalls Viktar. “However, as I started contacting factories, I realized they had nothing suitable for us.”
“The motors are all connected with a single cable, that’s why they have a hole in the middle,” Viktar describes the design of the servo motors that act as robot joints. “You can also connect the cable to a PC and then just command: motor 1, turn to such and such angle at such and such speed and acceleration. The motor will execute the commands with precision of 0.01 degree.”
The only purchased component in Rozum Robotics servos is the gearbox.
With “impressive power packed into a small size,” the Belarusian cobot weighs only 8 kg—3 kg less than the product by the company’s closest competitor. Only the Canadian robot from carbon fibre-reinforced plastic weighs less. The average weight of an industrial cobot ranges about 50 to 250 kg.
What are cobots and who makes them
Factories first started to use industrial robots in the 1960s, so the robotic technology is actually over 50 years old. Those were very complex mechanisms (it could take up to one and a half years to commission a robot) that posed danger to humans. Therefore, it was a common practice to “encage” them into so-called safety cells.
In 2005, the Danish “Universal Robots” company came up with a cobot, which is simple, cheap, safe, and easy-to-program. In 5 years, other major players in robotics presented their cobots. It took them almost 5 years to develop the solutions because big companies are really slow.
Today, major cobot producers include 5 giants, with 5 startups literally breathing down their neck. Being, on average, three years old, the startups are either at the stage of prototyping or just begin selling their products.
Every 2–3 months, one or two more manufacturers appear, mostly in Asia. In the next 10 years, the cobot market will become 100 times larger, as analysts predict. Whereas the market of conventional robots grows at 5 percent annually and its shares are essentially distributed among manufacturers, the more agile cobots are constantly conquering new niches.
Some forecast that cobots will take over more than half of the market, leaving it to the bigger robots to handle the heavy weights, such as automotive bodies. According to statistics, over 80 percent of cargo at manufacturing sites weigh less than 3 kg.
The team is looking for specialists who share “the passion”
Over a year and a half, the Rozum Robotics team has grown from 3 to 15 people. If the company continues as planned, the staff will include 25 people by the end of 2018 and 35 people by the end of 2019.
“In our company, there are specialists with experience of 15 years in microelectronics and two students who won prizes at specialized competitions,” says Viktar. “I met the students at a robotic exhibition, and they have never disappointed me since then. Though young, they have wits quicker and hands more skilled than most experienced specialists. For instance, it took them just a few days to program the joystick to move with a tenth-millimeter precision.”
The CEO believes that one more thing that unites the team is the common vision of how robots can change a human life.
“We’d like to hire people, not just talented, but also sharing our passion and willing to help the humankind to advance. To be more specific, we’d use good analysts with a technical set of mind who could resolve a system into components and then get it back together,” comments Viktar.
There’s no option mechanism implemented yet, but Viktar intends “to make an interesting offer” to its employees after the start of sales (not ICO, it seems).
Can Belarus catch up with the robotic trend?
For now, Belarus is a robot-free zone—the robots are not so numerous here.
“The global robotization level is 69 robots per 10,000 workers, but the figure is an average value, hardly reflecting all the variations. Thus, in South Korea, the value is 300, in Germany—250, and in Russia—it’s just one. There’s no such statistics for Belarus,” says Viktar. “In our country, most people still think: why purchase an expensive robot, if you can get a human to work for small money? The thing is that cheap labor force is no longer a competitive advantage. Can you guess what is the most promising market in terms of manufacturing automation? It’s China, where cheap labor is foundation of the national economy. However, it is clear for them now that the happy times are over, as both Europe and America are extensively adopting robots. Why get Chinese workers to cut metal with their rough hands, if you can do it with a cobot? Cobots can work anywhere—in China or America.”
The startupers believe Belarus should try to catch up with the trend, too. In spring, the Rozum Robotics team visited a number of Belarusian enterprises to see “whether robots can fit anywhere at all.”
“We saw enterprises where 10 people were doing what a single person could easily cope with. We’ve also been to branches of foreign plants, such as Coca Cola, that are automated shipshape. The plant itself is a gigantic robot: bottles are moved and filled with the drinkpotion automatically. Why can’t we have the same at our plants?” wonders Viktar.
Some Belarusian manufacturers came to the Rozum Robotics workshop on their own initiative, to test their cases in a laboratory.
“It was surprising that Belarusian companies were actually interested in robots. In the footwear manufacturing, for instance, there’s high personnel turnover in one of the routine operations—sole glueing,” comments the head of Rozum Robotics. “A robot doesn’t care if the glue smells, it needs no special air extraction, and the portions of glue it applies are always the same and precisely matched to avoid spills or soles coming off.”
Rozum Robotics has already received requests from an agricultural enterprise (greenhouses with no human attendance), a healthcare company (diagnosing bone fractures using ultrasound sensors instead of X-rays), and the media industry (a moving camera). An ordinary car enthusiast also contacted the company: he wants to build an automatic car wash in his garage “to get the car fresh in the morning.”
“We don’t really expect the Soviet-time industrial giants and even large 20-year-old companies will rush to buy a robot because of the hype,” explains Viktar. “Rather, we bank on “hardware” startups, especially where high precision is essential, such as in manufacturing of laser and medical equipment. We often hear them complain that it is difficult to assemble a prototype manually.”
Who will sign for a robot in the safety and health log?
According to the stratupers, the “slow” 8-kilogram robot can’t harm a human. Its maximum speed is less thanabout 2 m/s (it could have been faster, if not for the limits set up by the developers). In addition, the PULSE robot has an intelligent safety system.
“Every servo includes an encoder that senses the torque at which the motor rotates. Besides, there are special safety mechanisms. In our robot, it’s a special safety board that monitors each of its servo motors. In case of an accident, the board will stop the robot, switching it to the emergency mode,” says the company’s CEO.
Who will sign for robots in the health and safety log? “It’s grown enough to do this on its own,” jokes Viktar. “Joking aside, robots are machinery, so it is the integrator and the final point of contact who bears responsibility for it. Rozum Robotics plans to sell and install its first robots on its own, whereas the rest of the world works through dealers.”
Should workers fear unemployment because of robots?
The startupers estimate that the payback period for the PULSE robot is 6 months: average labour costs in Europe amount to 4,000 Euro a month, tax inclusive, which makes 24,000 Euro a year—more than the cobot costs.
“Since a cobot can work three shifts without stopping to have a smoke, you can triple the amount,” adds Viktar.
In Belarus and Russia, where the workforce is cheap, the payback period will sure be different. According to the team, workers are wary of robots, because they don’t know what to expect from them.
“Our mission is to tell people that cobots are not about unemployment, chaos, or Apocalypse, but rather about benefits,” says Andrei Shaurko. “It will let them engage in more complex and creative jobs, no round things to move or square things to roll.”
“A hundred years ago, about 90 percent of people were engaged in agriculture,” emphasises Viktar. “Were it not for the tractor invention, the Rozum Robotics team would probably be working in the fields right now.”
Will automation affect software engineers?
“Automation in IT is in full blast,” speculates Viktar. “Take, for example, quality assurance. Humans still write test cases, but they also use algorithms to run tests cyclically several times a day, which cope with the task better than humans. I am not sure when robots will actually start programming, but, in my opinion, automation will first affect those software engineers who build standard applications out of framework blocks. Accordingly, those who develop the framework blocks will be the last ones to be replaced by robots.”
The future is now
Viktar thinks we all live in a wonderful time: the hardware and software (the latest artificial intelligence solutions) will soon merge to get us a robot that we saw in books only. It is hard to tell yet whether the robots will be good or evil, as Elon Musk fears, but the humanity will sure find a way out.
The bookcase in the Rozum Robotics office is packed with books about robotics and cognitive psychology.
“This is my hobby—the science that studies how the human brain develops, from childhood to adulthood,” explains Viktar Khamianok. “How it learns to see things, to hear, to read, to understand, and think. When I look into the processes, I try to figure out how this could help us with our robot. Our robot already has a camera and can play checkers with a human, using a self-learning neural network. Our plan for the next 3–5 years is to learn him to see and think. I believe, we’ll finally switch to making smart robots, the ones that can do something beyond the exact commands.”