What projects does it make sense to start and what does the industry need from IT right now

One of the main tasks of IT in the manufacturing sector right now is to reduce the number of accidents.

What is changing in production now?

One of the most important issues now is zero mortality. Losses of personnel occur for two main reasons: non-compliance with safety regulations and alcoholic adventures with non-compliance with safety regulations. The second important issue is the control of the correct operation and the optimal load of production.



Everything goes to the point that in ten years all employees of the production will be “under the hood”. Now there are already pilots, but before the massive spread of technology a couple more years. There will be a production manager interface that looks suspiciously like a computer game. There, the position of each employee, the properties and conditions of each machine, the sequence of tasks, the well-being of everyone, the type of operations performed. The point is to digitize everything and control the correct execution of processes. And another auxiliary AI will optimize the tasks of employees - there are already examples where work is distributed according to the optimal loading of machines and the skills of workers, it is correctly clustered and built in the optimal sequence.



The latest leap in machine learning has made it possible to implement a ton of everything in industry. Even surprisingly old iron can be used in a new way, for example, searching for weapons in containers using an X-ray unit and a neural network for recognizing parts of rifles and pistols is no longer news. A lot of things are happening in the world, but right now we are mostly focusing on video analytics: from smart buses and trams that “knock” on all traffic violators around, to controlling production operations and expanding the intelligence of workers with a “line in the eye”. There are a lot of things about robots - and this is not about single UAVs, but about managing a single operator at once with a swarm. This is a real agricultural request from a Chinese customer, their economy does not converge without the ratio of “one person - five drones”.



Well, below are more examples of specifics and a proposal to participate in all this.



In general, I want to talk about the practical connection with Russian industry and specific tasks. If you are interested in talking with engineers and managers of the real production sector, then on November 16-17 in Moscow we will have a hackathon with real tasks from real industrial companies such as Sibur, Fosagro, UralChem, Gazprom Neft, and on real modern tools. Below is what is important now and to whom.



Now about security. Here is what the tools already have, but needs to be integrated into the realities of specific industries:

• Very high mortality during high-altitude work. A very real scenario - breathalyzers on cabinets with the same high-altitude equipment on the one hand, and tracking the Z-coordinate of the worker on the other. If it suddenly changes, but he didn’t take personal protective equipment, and he doesn’t have a permit, we can conclude that he climbed somewhere. Further, the intelligence of production immediately yells into his loudspeaker: "Ivanov, take a safety belt." The notification can be configured in accordance with the intuitive UX and to: "Ivanov, #% * @, where did you go, # @ # @?".
• Breathalyzers at all intermediate stages - in the cab of special equipment, when climbing into an electrical panel and so on. And then we have all come sober, but in time to manage to accept.
• Optimal reactions to the accident - while we are talking about individual routes to exit the danger zone, there will be scripts for coordination of workers to eliminate them.
• The connection of all sensors with ACS TP, when you can immediately determine the type of accident and its development scenarios. Well, just scare those who crawl close to a working machine at the wrong time.

Again, the hackathon has the task of optimizing the operation of the ventilation system - now it works at a nominal value of 100% of the time, and should be based on the presence of people, the types of work in the zone and the data of gas analyzers for an emergency.



Of course, in the new digital ecosystem, it will be easy to rewind visualization and investigate any incident. And both on the "raw" video, and in the "digital model" of production.



Separately, it should be said about predictive repairs - here we had a post about monitoring electric motors and generators by the noise that they give back to the power beam. This allows you to very accurately diagnose their condition and repair on time. Which is much cheaper. By the way, now Russian Railways already uses a system of predictive predictions of repair on wheelsets and at the same time fixes the document flow between contractors in the blockchain based on the Exonum platform (these tools can also be felt on a hackathon and written something for them).



Another thing about repairs is the systems for expanding the intelligence of workers. In a simple case, it looks like an assistant with augmented reality glasses, who will tell you how to make emergency repairs of expensive equipment right on the spot (if possible). Similar assistants are planned for paramedics and field surgeons in the future, but the difference between repairing equipment and repairing a person at about five years is likely. Training on the basis of augmented reality already exists - our CROC VR did such things for industrialists in terms of developing staff skills in working with equipment, its assembly, disassembly.



A combination of intelligence expansion and video analytics will also help to check the quality of technological processes and control the quality of parts. The worker took the part in his hands, his glasses were immediately measured, showed all the defects by the callouts from the part and said where to put it. Everything will be captured. A worker will be needed only as an avatar for artificial intelligence in one of the branches of development of industrial automation. But then he will be replaced by robots.

Robots

Drones are the second very promising area. It so happened that now we are talking only about flying drones, and most often in the form of quadrocopters and hexacopters. Let me remind you that there are still flying drones in the form of airplanes and cruise missiles, drones in the form of various animal-like forms (the Japanese snake looked promising for work in the rubble, but something has not been heard from them for a long time about the rapid development), all sorts of drones climbing the wires , rolling drones (we experiment a bit with supermarket trolley drones), floating drones, and “unkillable” drones. He will fight against the wall and nothing will happen to him. There are a lot of drones, and they are needed for everything from trivial fishing to concrete cases of deepening the fairway in ports where large vessels arrive. But all the same, it is flying robots that develop most rapidly.



We already wrote about mapping . Drones can do a lot in production, and pipelines can also be inspected. A lot of agricultural applications: from accurate spraying to crop accounting. There are automated harvesters.



Field for accidents and emergencies - we have already completed several pilot projects for the operation of drones in hazardous working conditions at production facilities, and the same Australian coastal service uses drones to dump drowning liferafts.



There is already a case when in one major quarry the miner tries to remove as many people as possible from the quarry itself. There is a constant danger of collision with heavy equipment and collapse of the walls of the quarry. Therefore, drones perform two tasks:

1. Help with the examination of the walls.
2. Used instead of people. Now experts come to the house near the edge of the quarry, launch drones from there and work from the remote control. Later, a drone station with interchangeable devices and chargers is planned there. The task is to remove people from the quarry while observing blasting operations.

A drone with a ground penetrating radar has already precisely replaced the manual skating of a cart when searching for cavities and large archaeological sites. But in the industrial and construction sectors, the question is in the interpretation of the data. Now expert people are doing this, and this is black magic. When we collect more data, we will train the neural network: for now, everything is based on the rules, which are still not fully understood. The expert says: here is the cavity, here is the underground river. And they believe him, because the only way to verify is to drill underground. The training sample is typed very slowly.

More promising areas

We have been dealing with the industrial topic for quite some time. If vectors are interesting, see Industry 4.0 ( or the Fourth Industrial Revolution ). We have worked out almost every aspect within the company in a special incubator for new projects. But specifically now I want to dwell on a couple of technologies that are needed right in the next six months by many.



With the advent of LoRaWan, you can play with very small low-current communication devices. The packages are small, so far they are calculated in kilobytes, but this is already real five kilometers from the worker’s helmet.



From what we know, now in trends there are different ways of identifying a person (silhouette and gait). This is necessary for a quick search - on one of the cameras it is identified by its face, and then the AI ​​leads it throughout the complex and writes its actions. By the way, a search in the crowd by the face has already been implemented on a tactical visor, the internal organs of many countries have it. Associated civil history - contactless payments based on the appearance of a person.



We are already making smart trains that themselves will catch hares based on this set of technologies.



Many are interested in controlling the driver’s behavior - this is not an autopilot, but tracking that you can’t drink, smoke, talk on the phone and don’t need to twist parts and equipment from the car, or take the load somewhere in the middle of the road. The latter are very interested in insurance companies, by the way.



There are already beacons on the end, which take the data of the registrars via Wi-Fi and analyze the driving style, violations and route.



For the controllers, a “who flew around” system is being made, it is similar to our old task - to show those with whom we greeted in the office today and no. We also make smart ships in the Middle East - there are control over the actions of employees on the ship during the transition, also unloading from cameras and analytics.



A very promising topic is automatic inspection and inspection of objects for damage (cracks, corrosion, and so on). It is used in factories and warehouses for incoming control of supplies. The ports already have cranes that do not take cargo until the drone flies over it and draws up a model that will be analyzed for damage. There are taps that do the same with lidars.



Actually clever observation of construction sites: how many materials were brought, which one went to, how many people really were at the construction site instead of the army prescribed in the contract of the general contractor.



In general, we are waiting for a brave new world, and we want to invite you to do it on our hackathon.

What exactly will be at the hackathon

Usually hackathons are more or less theoretical. But we saw what is right to do wrong, and immediately solve practical problems that seem to some to be the future. You need to get to MVP in two days, consulting with our experts in the course of work, and show it. In fact, this is the principle of our accelerator , in particular, to quickly reach a prototype.



This approach eliminates the main problem of hackathons related to the fact that a ready-made team comes there and shows their ready-made achievements.



Everything is different with us: practice at once, representatives of customers on the site and all the tools right on the spot. We assume that you will take ready-made libraries, dock ready-made modules and be able to prove that a specific problem is solved using IT tools here and now. For example, the task of creating a smart ventilation system to reduce energy costs can be solved using BLE (including) and using industrial Wi-Fi from Cisco . Hackathon participants will test the solutions in the CROC Cloud - it will be more than enough for all processes that require high performance. We have already participated with our team in such a banking hackathon in Oman, and there are all the tasks for a particular bank. They made a confirmation of the person’s identity from photographs of documents and selfies - and immediately received an international contract for the introduction of products within their infrastructure inside their integration with their Internet banking modules. It works something like this: first, the program determines that a living person is holding the phone - the task is given to smile, blink with his left eye or raise his right hand to the camera. There is a comparison of photos obtained as a result of the task and the person in the photo from the documents. Further voice identification. Further, without a visit to the bank, a person can open an account and gain access to it by his face and voice. All recognition takes place on the device in real time, after which the pre-processed data is sent to the bank's systems to run the corresponding checks. When you visit the bank, the system recognizes the person by the image from a regular camera and greets.



As a result, the situation is as follows:

1. Until September 30, we collect tasks from the industry and make an accurate list of representatives on the hackathon. If you are from industry and you have a task to study, send, here are the details .
2. Until October 25, you can apply for hackathon teams.
3. The final will be November 16-17 in Moscow.

Who needs this?

• You can try yourself in solving real industrial problems. You can break out of the routine for a couple of days and make a project. Well, or not.
• Find a team: we have a special VKontakte group for those who want to cooperate with someone in advance.
• Understanding what industry needs - it can be important for your future independent project.

Try vendor tools: we will have ready-made kits for deployment at the capacities of our cloud . For example, in the task of one metallurgical company, it will be possible to create blockchain solutions on the high-speed framework Exonum (Bitfury) , which recently entered the Russian software registry , and a top-secret Masterchain (blockchain platform developed at the site of the FinTech Association (AFT) together with key participants Russian financial market and the Bank of Russia), Cisco will bring tools for industrial Wi-Fi , building a LoRaWAN network, Cisco Kinetic IoT platform , Cisco VSM video surveillance .



What we want: to prove to the industry that it is possible to quickly solve specific problems, to hold the first hackathon with the integration of ready-made solutions.

Which expert will be?

Where to see the tasks?

Examples here on the site . Closer to the hackathon there will be more. All of them are extremely specific, for example, in the case of recognition of individual protective equipment, there will be exactly those that are used at the customer’s production site. It is possible that those who do not yet have industrial detectors will get there, for example, in our monitoring of personal protective equipment we did not give up a black cap-hat that is almost not recognized on brunettes.



We have experience in solving such problems, so we know what we need for this: access to tools, on-site experts (engineers and business people of our customers, our consultants) and pizza. I would love to talk about our ecosystem for solving production problems, but this is rather a topic for a separate post. For now, briefly - we help them land the specific needs of the plant or conceptual stories on technology. That is, we show the benefits of "before" and "after." What exactly to receive, what money to save, how much to earn, how many lives to save. Second, we can allow ourselves to test hypotheses such as spending 500 thousand rubles to see if a video analyst can catch a theft on a truck. We make a model, show the economic effect and then the customer makes a decision on implementation. We also know how to travel to specific enterprises and do everything there until the result. For example, recently there was a case in Siberia where we introduced industrial vision. They fumbled with cameras, then with frosts, then with launches of equipment for a couple of megawatts, then they rolled the product up to these conditions, then they slightly changed the infrastructure of the plant itself (there were no strong admins) and only then it worked. We offer the same thing, but so far without the romance of going to polar bears. To do the real thing. As much as possible.

References

• Industry 4.0 .
• Recognition of PPE .
• Smart helmet .
• Hackathon site and all the details .
• Examples of tasks on the hackathon site .
• VKontakte group to search for teams .
• My mail for any questions is sstrelkov@croc.ru.