Every year, rescuers search for tens of thousands of people missing in the wild. From the cities, our technological power seems so huge that it can handle any task. Like, take a dozen drones, hang a camera and a thermal imager on each, screw on the neural network and that's it - it will find anyone in 15 minutes. But this is not true at all.
Until now, technology has run into a lot of restrictions, and rescue teams are combing huge areas with hundreds of volunteers.
Last year, the Sistema charity fund launched the Odyssey project to find new technologies for finding people. Hundreds of engineers and designers took part in it. But even technically savvy and experienced people sometimes did not suspect how impenetrable the forest was for technology.
In 2013, two little girls, Alina Ivanova and Ayana Vinokurova, disappeared in the village of Sinsk in Yakutia. Colossal forces were thrown into their search: a hundred volunteers, rescue teams, divers, drones with thermal imagers were equipped. Helicopter footage was posted to the public so that everyone could view the recordings on the Internet. But the strength was not enough. What happened to the girls is still unknown.
Yakutia is huge. If it were a state, it would be among the ten largest in terms of area. But less than a million people live on a gigantic territory. In such an endless deserted taiga, Nikolai Nakhodkin worked for 12 years in the Rescue Service of Yakutia, 9 of which - as a leader. When the conditions are worse than ever, and there are few resources, you have to invent new ways to find people. And as Nikolai says, ideas do not come from a good life.
Nikolai Nakhodkin
Since 2010, the Yakutia Rescue Service has been using drones. This is a separate organization from the Ministry of Emergency Situations of the Russian Federation, funded by the republic itself. There are no such strict regulations for equipment, so drones began to be used in the Ministry of Emergency Situations much later. Also in the service there is a scientific group, where enthusiastic engineers are developing applied technologies for rescuers.
“The existing search methods that the Ministry of Emergency Situations, the rescue services, and all law enforcement agencies have have not changed since the 30s. The tracker follows the trail, the dog helps not to stray,” says Alexander Aitov, who was the head of the scientific group. “If a person is not found, a whole village rises in Yakutia, two, three. Everyone unites and combs the forests. To search for a living person, every hour is important, and time goes by quickly, quickly. It is always scarce. When the tragedy happened in Sinsk, a lot of people and equipment were involved, but without result. Similar situations occur in searches in the deserted taiga. In order to somehow fix this, the idea came not to perceive the missing person as a passive link, but to use his own desire to escape and an active thirst for life.
Rescuers-engineers decided to assemble rescue light and sound beacons - rather large, but lightweight devices that emit a loud sound and glow for a long time, attracting attention day and night. A lost person, having come to them, will find water, biscuits and matches - and at the same time instructions to sit still and wait for rescuers.
Such beacons stand at a distance of three kilometers from each other and encircle the approximate search area for the missing. They make a low sound, as if a car is roaring - because high frequencies are much worse distributed in the forest. Often the rescued thought they were walking towards the sound of a road or a company of tourists about to leave.
Lighthouses were incredibly simple. This is not the first time the scientific group has implemented elementary, but ingenious solutions.
“There, for example, they developed a floating suit for rescuers. Pants and a jacket look like ordinary overalls, but in the water they keep a person afloat. Speaking quite utilitarian, the suit is two-layered. Polyurethane foam granules are sewn inside. There is a development for diving a diver at low temperatures. When the compressed air expands in the cold, the valves are covered with frost, and the person suffocates. Several institutes could not figure out what to do with it - they developed special materials, made electric heating, introduced all sorts of modern approaches.
Our guys solved the problem for 500 rubles. The cold air that comes from the cylinder (and they go under water at -57) they passed through a coil passed through a Chinese thermos. The air heats up, people go under water and can work there.”
But the beacons are too simple, they lacked many useful functions. During search operations, the rescuer regularly had to run huge distances to check each beacon. If there are ten beacons, then the rescuer must walk 30 km through the taiga every 3-4 hours.
In 2018, the Sistema Charitable Foundation launched the Odyssey project, a competition for teams that, using new technologies, will try to find the latest ways to save missing people in the wild. Nikolai Nakhodkin and Alexander Aitov and their friends decided to take part - they called the team "Nakhodka" and brought their simplest device to improve it in competition with others.
According to the Ministry of Internal Affairs, almost 2017 people went missing in Russia in 84, and half of them were not found. An average of XNUMX people searched for each missing person. Therefore, the mission of the Odyssey contest was “to create technologies that will help find missing people in the forest without a source of communication. It can be instruments, sensors, drones, new means of communication and everything that your imagination is capable of.”
“Of non-obvious solutions - or fantasy ones - one can name an airship equipped with a bioradar system. But the team did not have a prototype, and they limited themselves to just presenting their idea,” says competition expert Maxim Chizhov.
Another team decided to use a seismic sensor, a device that, among the vibrations on the ground, can recognize human steps and show the direction from which they are heard. With the help of the prototype, they even managed to find an extra who portrayed the “lost” (as the participants affectionately called them), but the team did not go far in the competition.
By June 2019, after several practice trials in the forests of the Leningrad, Moscow and Kaluga regions, the 19 best teams reached the semi-finals. They were given the task of finding two extras in less than 2 hours on a plot of 4 square kilometers. One moved through the forest, the other lay in one place. Each team had two attempts to find a person.
“Among the semi-finalists, one team wanted to create a swarm of drones that would fly under the canopy of trees, controlled by artificial intelligence, determining the direction of movement, flying around trunks, dodging branches and branches. Using AI, he would analyze the environment and identify the person.
But this solution is still quite far from the implementation in a functioning form. I think it will take about a year for it to work, at least in test conditions,” says Maxim Chizhov.
The ALB-search team was close to success. They had a loudspeaker on board that connected to a walkie-talkie, a microphone that could listen to the surrounding space, a camera and a computer with AI and a trained neural network that processed images from the camera in real time, where a person could light up.
“The operator could analyze not thousands of images, which is physically impossible, but dozens or even units, and then make a decision: whether to change the drone’s route, whether an additional drone is needed for reconnaissance, or immediately send a search group.”
But most teams faced similar problems - the technology was not adapted to the conditions of the real forest.
The computer vision that many relied on worked when tested in parks and woodlands - but proved useless in dense forest.
Thermal cameras, which about a third of the teams hoped for, also turned out to be ineffective. In the summer - and this is the time when most people disappear - the foliage heats up so that it turns into a continuous hot spot. It is easier to search in a small gap at night, but there are still a lot of heat spots - heated stumps, animals and much more. A camera could help to verify suspicious places, but it's of little use at night.
In addition, thermal imagers proved to be difficult to obtain. “Unfortunately, due to the restrictions imposed on us by the EU and other countries, good thermal imagers are not available in Russia,” said Alexei Grishaev from the Vershina team, which relied on this technology.
“Thermal cameras available on the market have a digital output at 5-6 frames per second and an additional analog video output with a high frame rate but poor image quality. As a result, we found a very good Chinese thermal imager. We can say that we were lucky - in Moscow there was only one such. But he gave out a picture on a small monitor, where nothing was visible.
Most teams used the video output. Our team was able to refine the model and get a high-quality digital image from it at a frequency of 30 frames per second. The result was a thermal imager of a very serious level. Better, probably, only military models.
But even these problems are only the beginning. In the short time that the UAV flew over the search area, cameras and thermal imagers collected tens of thousands of images. It was impossible to transfer them to a point on the fly - there was no Internet or cellular communication over the forest. Therefore, the drone returned to the point, records were downloaded from its carriers, spending at least half an hour on it, and as a result, they received such an amount of material that it was physically impossible to watch even in hours. In this situation, the Vershina team used a special algorithm that highlighted images where thermal anomalies were detected. This reduced the processing time.
“We saw that not all the teams that came to the qualification tests understood what the forest was. That the radio signal propagates in a different way in the forest and is lost rather quickly,” Maxim Chizhov announced at a press conference. “We also saw the surprise of the teams when the connection disappeared already at a distance of one and a half kilometers from the starting point. For some, it was a surprise that there was no Internet over the forest. But this is reality. This is the forest where people get lost.”
The technology based on sound and light beacons showed itself well. Four teams reached the final, three of which relied on this decision. Among them - and "Nakhodka" from Yakutia.
“When we saw this jungle near Moscow, we immediately understood that there was nothing to do with drones there. Each tool is needed for its own task, and they are good for inspecting large open spaces,” says Alexander Aitov.
By the semi-finals, there were only three people in the team who went through the forest on foot and placed beacons in the search area. And while many were solving engineering problems, Nakhodka worked like rescuers. “It is necessary to apply a non-agrarian psychology when you simply cover the area. You have to behave like a rescuer, put yourself in the place of the missing person, look at the approximate direction where he can go, what paths.
But the lighthouses of "Nakhodka" by this moment were no longer as simple as several years ago in Yakutia. With the help of Sistema grants, the team's engineers developed the radio technology. Now, finding a lighthouse, a person presses a button, the rescuers immediately receive a signal and know exactly at which lighthouse the lost person will be waiting for them. The UAV is needed not for searching, but in order to lift the radio signal repeater into the air and increase the radius of transmission of the activation signal from the beacons.
Two more teams developed entire search complexes based on sound beacons. For example, the MMS Rescue team has created a network of portable beacons, where each beacon is a repeater, which makes it possible to transmit a signal about its activation even in the absence of direct radio communication with the search headquarters.
“We have a group of enthusiasts who have taken on such a task for the first time,” they say. “We were engaged in other industries - technology, IT, we have specialists from the space industry. We got together, stormed and decided to make such a decision. The main criteria were low cost and ease of use. So that people without training can take and apply.
Another team - "Stratonauts" - with the help of a similar solution was able to find an extra faster than anyone else. They developed a special application that tracked the position of the drone, the location of the beacons, and the position of all rescuers. The drone that delivered the beacons also acted as a repeater for the entire system so that the signal from the beacons would not be lost in the forest.
“It wasn't easy. One day we got really wet. Two of our people went into the forest through the windbreak, and they realized that this was far from going on a picnic. But tired and satisfied, they returned - after all, we found a person in both attempts in just 45 minutes, ”says Stanislav Yurchenko from the Stratonauts.
“We moved the beacons to the center of the zone with the help of drones to provide the greatest coverage. During one flight, the drone can transfer one beacon. It's long - but faster than a man. We used small compact DJI Mavick drones - one beacon the size of it. This is the maximum that he can carry, but it turns out to be budgetary. Of course, I would like to find a completely standalone solution. With AI to have the drone scan the forest and identify drop points. We now have an operator, and after a kilometer, if you do not use additional devices, the connection ends. Therefore, in the next stage, we will come up with something.”
But not a single team found an immobilized person, and most importantly, they never figured out how to do it. Theoretically, only the Vershina team had a chance to find him, which, despite all the difficulties, was able to find a person and go to the final with the help of a thermal imager and a camera.
“Initially, we had the idea to use two aircraft-type drones,” says Aleksey Grishaev from Vershina, “We developed them to determine the composition of the atmosphere, and we still have the task of making an all-weather UAV. We decided to try them in this contest as well. The speed of each is from 90 to 260 km / h. The high speed and unique aerodynamic characteristics of the UAV provide the ability to search in all weather conditions and allow you to quickly scan a given area.”
The advantage of such devices is that they do not fall when the engine is turned off, but continue to glide and land on a parachute. The downside is that they are not as maneuverable as quadrocopters.
On the main drone of the Vershina, a thermal imager and a high-resolution camera modified by the team are installed, on the second, only a camera. On board the main UAV there is a microcomputer that, using the software developed by the team, independently detects thermal anomalies and sends their coordinates with a detailed image from both cameras. “So we don’t have to watch all the material live, and this, for understanding, is about 12 shots per hour of flight.”
But the team created the technology of the aircraft recently, and there were still many problems with it - with the launch system, with a parachute, with an autopilot. “We were afraid to take him to the test - he could just fall off. I wanted to avoid technical problems. Therefore, we took the classic solution - DJI Matrice 600 Pro.
Despite all the difficulties, due to which many abandoned cameras and thermal imagers, Vershina was able to find an extra. This required a lot of work, firstly with a thermal imager, and secondly with the search methods themselves.
For three months, the team tested the technology, which allowed the thermal imager to view the ground between the canopies. “There was some luck, because the route of the extras ran through such forests that not a single thermal imager would see anything. And if a person is tired and sits somewhere under a Christmas tree, it will be unrealistic to find him.
From the very beginning, we refused to completely comb the forest with our UAVs. Instead, we decided to search for a person by flying over clearings, clearings and open areas. I arrived at the site in advance to study the area, and using all the maps available online, I drew UAV routes only over those places where a person could theoretically be seen.”
According to Alexei, it is very expensive to use several drones at once in a bunch (one carrier with a technical solution for searching on board costs more than 2 million rubles), but in the final it will be necessary. He believes that this gives a chance to detect a motionless extra. “We initially wanted to look for a lying person. It seemed to us that we would find a mobile one anyway. And the teams with beacons - they were looking only for the moving one.
I asked Alexander Aitov from the Nakhodka team - doesn’t it seem to them that everyone has already buried a static person in advance? After all, beacons are useless for him.
He considered. It seemed to me that all the other teams were talking with smiles and a twinkle in their eyes about solving engineering problems. The guys from MMS Rescue joked that a dropped beacon could fall right on a lying person. The "Stratonauts" admitted that this is the most difficult task for which there are no ideas yet. And the rescuer from Nakhodka spoke, it seemed to me, with a mixture of sadness and hope:
- We have lost a girl in the taiga, at the age of three and a half. She spent twelve days there, and for ten days a large number of people searched. When they found her, she was lying in the grass, from above she was practically invisible. Found only by combing.
If beacons were placed ... at three and a half years the child is already quite conscious. And maybe she would have walked up to him and pressed the button. I think some lives would have been saved.
- Did they save her?
- Her yes.
In the fall, the four remaining teams will go to the Vologda Oblast, and the task before them will be much more difficult - to find a person in a zone with a radius of 10 kilometers. That is, on an area of more than 300 square kilometers. In conditions where the drone has half an hour of flight, vision breaks into crowns, and communication disappears after a kilometer. According to Maxim Chizhov, so far not a single prototype is ready for such conditions, although he believes that everyone has a chance. Grigory Sergeev, chairman of the Liza Alert search and rescue team, adds:
“Today we are ready to use a couple of technologies that we have seen, and it will be effective. And I urge all participants and non-participants - guys, test technologies! Come find us! And then it will not be a secret for anyone that the forest is opaque for the radio signal, and the thermal imager does not see through the crowns. My main dream is to find more people with less effort.”
Source: habr.com