Shulou(Shulou.com)06/02 Report--

The five-hour fire destroyed the 850-year-old Notre Dame Cathedral in Paris. In this moment of regret for all mankind, we cannot forget either:

Founded in 1284, the Yuan Dynasty Guoguo Temple has been plagued by fires for many times; in 2010, the century-old school of Tsinghua University was completely destroyed by a fire; in 2014, Glasgow College of Art and countless art treasures in Scotland were completely destroyed; in 2017, Lingguan Building, Asia's tallest wooden tower in Sichuan, was finally burned down due to a fire in the hall; In 2018, Brazil's National Museum set fire to 20 million collections, including Brazil's oldest human fossils.

And those ancient buildings that survived temporarily could not rest easy, but lived in the anxiety of fire all year round. After Brazil's museums were burned out, our country quickly organized a nationwide fire safety inspection of museums and heritage buildings. As soon as the news of the fire at Notre Dame de Paris came out, the Forbidden City held an emergency fire prevention mobilization meeting; and the appeal for preventive protection of the tallest existing all-wood building in Ying County received widespread attention and support.

The threat of fire to important cultural buildings hovers over human civilization like a ghost all year round. The reason why every fire makes people extremely sorry and worried is that on the one hand, the cost is too high, the damage caused by fire to human architectural art culture is almost destructive and irreparable; in addition, many ancient buildings often have special structural techniques and materials, which makes many modern fire protection measures seem a little weak.

For example, foam extinguishers often used in fire rescue may cause secondary damage to fragile wooden structures in traditional buildings; the geographical location and surrounding infrastructure of some buildings are relatively old, and large-scale fire-fighting facilities such as fire engines cannot enter on a large scale, etc.

In this civilized defense war between human beings and "fire dragon," predecessors have summarized and explored many methodologies, such as intelligent fire hydrant, automatic alarm system and so on. So, when the baton of technology is handed to AI, can it do more? The answer has to be yes.

AI challenge fire, what core combat power?

Cutting-edge technology, represented by AI, is being placed on fire rescue missions. The reason why it can be a big task may be attributed to the three core combat capabilities of AI:

1. Deep learning + intelligent data terminal combination to predict and determine the fire risk and the specific circumstances of the accident, from the source to control the fire;

2. The cloud dispatching of smart cities can quickly formulate disaster relief plans, help firefighters and related materials to enter the site efficiently, and buy time for rescue work.

3. Intelligent robots and the like can accomplish tasks that humans and traditional equipment cannot, such as heading to heat resistance.

Of course, such a general generalization may still be a bit vague. After all, the fire prevention project of key buildings is a huge comprehensive system including prevention, disaster relief, repair and reconstruction, etc., during which various special difficulties may be encountered.

In order to fully understand the specific role and potential value of AI in important building fire rescue, we try to restore how AI technology clusters work comprehensively in fires through several key scenarios.

Scenario 1: Fire Prevention

For ancient buildings such as Notre Dame de Paris, Ying County Wooden Pagoda and Forbidden City, prevention is always the first priority. However, depending on manual inspection, it is difficult to ensure the timely detection of fire due to the influence of complex structure and scale.

Therefore, it is necessary to cooperate with the technical solution of fire prevention through cloud network + intelligent IoT"software and hardware."

Let's start with hardware. Many teams are already experimenting with placing IoT devices with wireless sensors in vulnerable locations. These devices monitor data in real time and upload it to the cloud, triggering an alarm mechanism in the event of an anomaly. Of course, the equipment also has an automatic repair function to ensure long-term stability and availability.

For some fires caused by non-flammable substances, such as line ignition or use of electrical equipment and other hidden fire risks,"current fingerprint identification" technology can be timely judged. The working principle behind it is a bit like Face Recognition. By extracting "current feature data" and judging whether the working state of relevant circuit equipment is abnormal, risk early warning can be realized.

You may have discovered that IoT intelligent fire equipment only exists as a data terminal, and the specific processing effect depends on the cloud platform and algorithm to guarantee it.

The overall architecture of the intelligent fire cloud platform is roughly divided into three layers:

The first layer: the device sensing layer, which is what we just mentioned, including smoke, temperature, inductance, smart camera and other sensing devices. For example, Huawei has already practiced narrowband Internet of Things technology in fire hydrant monitoring, smoke detection and other fields in 2014. BAT has also cooperated with many provincial and municipal governments to deploy urban fire sensors and edge computing gateways.

Layer 2: Cloud PaaS Layer. The data collected by the sensors is aggregated to the cloud for processing, and the predictive power of AI also comes into play here. But objectively speaking, there is no mature algorithm to detect fire location and potential risk effectively.

On the one hand, the system cannot be fully trained due to the insufficient data scale supported by the incident; on the other hand, the data returned by the terminal IoT and the geosynchronous operating environment satellite (GOES) has a certain delay, such as several minutes to provide high-resolution images, which greatly reduces the timeliness of prediction. And the algorithms aren't exactly accurate. Researchers at the University of California, Davis, have devised ways to detect fires, some of which detect backyard campfires and barbecues.

At present, it may take years for neural networks + deep learning to predict fires with a certain accuracy.

Level 3: Application SaaS layer. The cloud prediction results will finally be presented to professionals through APP, web and other application platforms, and fire prevention monitoring and early warning, fire material control and other business work will be completed.

Although we want to "prevent" major disasters, AI algorithms seem to be helpless for fires with insufficient historical data. However, the intelligent upgrade of hardware obviously has extremely important practical significance and feasibility for disaster prevention and relief.

Scenario 2: Fire Rescue

Relying on manual inspection and AI prediction to prevent fire obviously has its own limitations and cannot achieve the real ideal effect. A California firefighter is even more blunt, arguing that one of the state's most valuable fire-fighting tools is actually the humble mobile phone, which allows ordinary people to call emergency services when they spot a fire.

If there is an unfortunate fire, the efficiency advantages and special abilities of cutting-edge technologies such as AI can really help.

The core of AI application in disaster relief scenarios lies in using technology to improve rescue efficiency.

First of all, in terms of scheduling, relying on the big data of smart cities, we can judge and decide the best rescue strategy in time.

For example, in route planning, intelligent evacuation can be realized according to the fire alarm position, people flow and traffic flow conditions, so as to save time for rushing to the scene. Rescue facilities are determined based on real-time data from satellites, smart terminals, city cameras, etc.

The intelligent fire truck is also equipped with GPS satellite positioning autonomous navigator. When receiving the alarm, it can display the alarm location, route, user name, etc., call up the fire fighting plan data of the rescue object, and plan the best driving route to the fire alarm point.

At present, with the efforts of science and technology companies and relevant departments, the early dispatch of emergency rescue and disaster relief has been effectively realized with the help of smart city network. For example, Shenzhen's "intelligent traffic platform" can quickly and automatically generate the shortest navigation route to turn all necessary roads into green lights when a fire is discovered, so as to ensure the rapid passage of fire vehicles, etc.

At disaster relief sites, intelligent technology can also greatly improve rescue efficiency. For example, relying on mature UAV equipment, in some rugged roads in Laocheng District, the first UAV with thermal imaging function can provide the most timely fire reconnaissance, fire source positioning and other key information. Combined with the fire cloud, it can also predict the spread path of the fire, adjust the strategy and divert the crowd in time to reduce potential casualties.

In addition, AI's cloud capabilities can help professional firefighters make more reasonable judgments.

As mentioned earlier, important ancient buildings often require special fire strategies to deal with. Take Notre Dame de Paris as an example, in addition to firefighters, technical guidance from medieval architectural experts is needed to formulate reasonable rescue plans according to the specific conditions of cultural relics to minimize losses. Notre Dame's fire plan abandoned the roof, which could not be saved, and devoted all measures to the protection of the internal structure, resulting in much less damage than expected.

Just think, if it is other non-famous buildings, if firefighters can obtain relevant information in time through digital data during rescue, can they avoid some losses that could have been avoided?

At present, many organizations are trying to use digital methods to retain important building information. For example, Google Arts & Culture has restored museums around the world to high quality on the Internet, and the database of Chinese ancient architectural cultural resources has also collected information from hundreds of national treasure-level ancient wooden structures through digital scanning and fidelity photography. In the future, through intelligent integration of these data and reasonable arrangement of disaster relief methods, front-line firefighters will be able to quickly make more targeted choices at the rescue scene, avoiding secondary losses caused by inappropriate methods.

This kind of system has been gradually constructed, and with the introduction of AI ability in the future, it will inevitably become an important yardstick to guide the rescue of ancient buildings.

In addition to firefighters, we can see a lot of intelligent robots participating in today's fire rescue scene. The value of their existence lies in their ability to help firefighters complete tasks such as dangerous environment detection and special space rescue.

The Stanford team, for example, has designed an "inflatable robotic snake" that can stretch its length freely. It is also equipped with autonomous computing chips and motion sensors that allow it to intelligently judge "road conditions." If you need to turn around an obstacle, the brain of the "air snake" sends a signal to the mechanical pump at the back end to achieve complex movements such as turning by filling different amounts of air on different sides.

Therefore, this air snake can accurately pass through obstacles to reach the fire. For church domes, caisson wells and other inaccessible corners,"air snakes" can replace humans to complete the fire. Due to its strong weighing ability, it can also help people who are pressed under falling objects escape from heavy objects of at least 100 kilograms.

Of course, this flexible robot also faces many problems. For example, the degree of intelligence is not high. Under the premise of lack of real command, only a small amount of "intelligent" reactions can be made by sensors. In essence, it is relatively mechanical and can only complete single rescue tasks. Soft materials, such as plastics, also perform poorly in heat resistance, and researchers are said to be considering changing from "air snakes" to hydraulic ones, but overall their resistance is relatively limited.

In order to improve the intelligence and adaptability of disaster relief robots, it also needs comprehensive progress in chip hardware, materials science, Internet of Things, intelligent algorithms and other technologies. In a short period of time, disaster relief may depend on human beings themselves.

Scenario 3: Post-disaster reconstruction

Because of the particularity of structure and material, it is almost difficult for fire to be "zero loss" for important cultural buildings.

Just now, French media announced the damage to Notre Dame de Paris, although lighter than expected, the main structure was preserved. But the spire had collapsed, and the upper half of the left tower and the famous rose window had been destroyed. The art works in the museum were also transferred and preserved.

The outcome of the disaster is regrettable, but how to rebuild civilization after the disaster is the key. Among them, the application of modern technologies such as 3D laser scanning has played an important reference value.

In 2015, art historian Andrew Talon's laser scan of Notre Dame became the most reliable reference for its post-disaster restoration. Talon's work left more than a billion data points, covering more than 50 locations inside and outside the cathedral, with color data accurate to five millimeters. The resulting photos are also very accurate.

Finally, the collected data is spliced into a "point cloud" of 1 billion points to create a realistic three-dimensional model. This data will directly assist in the restoration of the church and restore it to its original appearance before the disaster.

There are many stories about smart technology intervening in fires that cannot be told one by one. In the face of such a powerful natural force as fire, human beings raised their "Dragon Slaying Knife" and launched a charge again and again. The end of the battle may not be victory, but it is definitely worth everyone's expectation and effort.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.