Shao Tianlan told us that from his visits to hundreds of factories, there are still a lot of manufacturing processes that have not been automated.
"We are the party B of the Party B," Shao Tianlan said. In his opinion, the industrial processes are numerous and complex, and the demand for automation is different, which also leads to the customization. The high investment in the early stage makes automation and intelligence become the exclusive of large industries and enterprises.
Mercamander's idea of how to get out of the "industrial customisation" dilemma is to leave the complexity to software. They make AI+3D vision technology become the brain and eye of the robot, make standardized hardware together, but also complete the feeding, testing, screw, cutting, welding and other complex work.
'Automation is far away from the tech community and investors, but in the last two years, it's clear that people are paying attention to new innovations in the old industry of industrial automation,' Ms. Shao said. In June 2022, Mercamander closed a Series C+ funding round.
Sales of industrial robots have increased tenfold in the past decade. Industrial automation is becoming an affordable technology dividend for small businesses. It is a vast, potential market, but also a place waiting for innovation, waiting for talent.
The following is a transcript of a speech by Shao Tianlan, founder of Mercamander, compiled by Geek Park:
I came back from Germany six years ago to start Mercamander. Mercamander is very much like an industrial autonomous driving company, except instead of driving a car, we drive a variety of devices, and we work on making industrial robots smarter through sensing and planning technology.
In the direction of AI+ industrial robot, we are one of the enterprises with the largest number of cases and the largest amount of financing in the world. We have AI, robot, sensing and perception technology, but the track we are in is automation, or industrial automation.
Is there anyone here who comes from manufacturing or industrial automation? Very few. Let me first share with you a number, which I think is amazing, which is 0.
It's an honor to be a part of this event. I did my homework before I came here, looking at all the speakers at GeekPark Innovation Conference over the past five years to see who's main business is industrial automation, and I got zero. Today it's 1, because I work in industrial automation.
Shao Tianlan, founder of Mercamander, delivers a keynote speech at IF 2023, an innovation conference at Geek Park
Over the past few years, there have been a lot of great corporate guests, including the Internet, consumer electronics, automobiles, cloud computing, etc., but industrial automation has been a little far from the tech scene.
Why am I here as an automation practitioner? Because it has a huge market, a hundred billion dollar fortune 500 company like Siemens, it's a huge industry. Of course there are other huge industries, such as hot pot, but there is not much innovation in these industries. There is a lot of innovation going on in the industrial automation industry, and innovation is rapidly changing an industry that doesn't sound that trendy or fashionable.
I'll put a number on the rapid change that innovation brings, which is 10 times. Sales of industrial robots have increased tenfold in the last decade, and the Internet may feel tenfold in 10 years is nothing, but for automation it is great. Today I'd like to share with you some of our observations.
01
The "fake" industrial automation boom
The automation circle is far away from the science and technology circle and investors. In the past six years, I have often heard about some problems. For example, car factories such as Volkswagen and Toyota often employ hundreds of thousands of people, and household appliance companies such as Gree and Haier are the same. The same data shows that there are still a lot of people working in manufacturing.
Others ask, in Germany, Japan and other countries, such a simple job has been automated? If so, their factories will not move to East and Southeast Asia. There is also a common saying that today's factories are advanced. For this speech, I went to find some materials. I searched the factory. In fact, many videos in the factory were cut out parts with relatively high degree of automation. I myself have been to hundreds of factories where there is no automation.
You see a lot of videos of, like, fully automated potato chip factories, where potatoes come in and chips go out in packages. These videos will give many outside the circle the impression that manufacturing is highly automated. But this kind of video is only a very small part of the manufacturing industry.
What does mainstream manufacturing really look like? I'll give you some examples. You have rows and rows of tens of thousands of people who are putting together small appliances like coffee machines by hand, and they have to do a lot of things. This is a foreign garment factory, there are also a lot of workers, in hand sewing and other work. That's typical of manufacturing, which still has a lot of non-automation.
What kind of manufacturing industry, or what kind of links, is its degree of automation relatively high?
The most typical is the stamping, welding, painting and other links in the automobile main engine factory, which has a high degree of automation, but also a large number of robot equipment is used. And things like chips, like fabs or the semiconductor industry, which are highly automated.
Standardised foods, potato chips, cokes, cans, etc., will also be highly automated. In addition, since the industrial revolution, the degree of automation in the textile industry is also quite high. There are shampoo such as daily, large amount of cosmetics, including a variety of drugs.
Beyond those, almost every other industry and process is much less automated.
In industries with high automation, we can find some rules and find that:
First, big business dominates. Like a pharmaceutical company. Pharmaceutical companies make a lot of money, car companies make a lot of money, billions of dollars.
Second, the production scale is large. Whether it's Coke, potato chips, or cars, it does it on a massive scale, all year round.
Third, the product iteration cycle is long. For example, Coke, in addition to the outer packaging to change the festival, most of the things inside, change is very slow. And cars, they may change faster now, but they used to take longer to produce.
Fourth, the process is stable. Beer, for example, may not have changed much in the process of beer for decades.
Fifth, the raw materials and production environment can be controlled.
02
Do the standardization program,
Let complex things move to software
Industrial automation is a very hard business, and we consider ourselves Party B's Party B. The demand for industrial automation is very much also very complex, thousands of industries and there are a lot of processes, each process needs different automation.
The previous automatic combination of sensors, motor these hardware, there is a high degree of non-standardization, environmental transformation, low flexibility. Low flexibility means that if the customer produces the product today, your automation scheme is OK, but if his product demand changes tomorrow, the automation scheme will no longer be appropriate.
We want it to be highly standardized and highly intelligent no matter what the requirements are. This intelligence is best and human is the same, do not need to design for the environment, the original human do, the robot in the past will do the same. At the same time, it is easy to deploy, and the best customer needs automation, send a robot over, in the afternoon can let it directly dry.
Shao Tianlan, founder of Mercamander, delivers a keynote speech at IF 2023, an innovation conference at Geek Park
We talked about it for a long time, but we couldn't do it because it was technically challenging.
Human beings are much better than robots in several points, one is the actuator. I'm not a very strong person, but it's easy for me to carry 20 or 30 kilograms, but that kind of energy density is unimaginable for current robot motors, one order of magnitude less. For example, my hand is not very strong, but it has more than 20 joints and a lot of muscles. Such degree of integration and flexibility is also beyond the capabilities of today's robots. Moreover, our brain is a very good control system. Such complex control ability is also difficult for today's robots to achieve. Even if you want to break through, the cost will be very high.
But we want automation to be better, so what can we do? You can move complexity from hardware to software as much as possible. Although we can't do human displacement like humanoid robots, but still can make industry more standard, more intelligent, more agile, more flexible.
03
Industrial automation can be "universal"
Let me give you some examples of how we have deployed products over the years, such as the most common process of feeding materials in manufacturing.
There are a lot of machines in the manufacturing industry, such as machine tools, hot processing, a variety of testing equipment, etc., each of which has a feeding demand, especially for discrete manufacturing factories. In many factories, workers stand in front of a machine with a box of semi-finished products from a previous process, which they need to put in a specific place. But we can also solve this problem through AI+3D vision technology, using standardized hardware.
The robotic arm is this kind of standardized hardware, which is shipped in bulk from the factory. Instead of making a specific mechanical design for the workpiece, we let the robot know how to pick up an object and put it in the frame through a visual +AI approach.
In our lab, we randomly placed a large number of objects in a box that you would find in a supermarket, in a variety of sizes and materials, and the task of the robot was to take these objects out one by one, just like a supermarket cashier.
In addition to feeding, there is also gluing. Thousands of different kinds of Windows and panels need to be gummed, and we use standardized robots and standardized vision systems that allow robots to identify and locate different Windows and panels and then glue them. The robot is a completely standardized product, so is the vision, and the conveyor belt doesn't have any special mechanical design for the workpiece, so the complexity of it, the trajectory of the robot's movement, is completely handled by software. For tires, now you put in a standardized robot, you put in a standardized system, you can track the movement of the car.
Today's robots can solve highly complex and non-standardized automation requirements through advanced vision, through artificial intelligence technology, through very standardized hardware, namely robotic arms.
Here is a video to introduce our products. Through optical design, we can enable robots to carry out very fine 3D scanning, and then complete many links with highly standardized hardware and software, such as feeding, testing, screwing, cutting, welding and so on.
In the past, industrial automation was more exclusive to a few large industries and enterprises, such as automobiles and chips, but through intelligent technology, we can let more industries, and a lot of small and medium-sized enterprises, can enjoy the automation technology.