Why don't airplanes fly by themselves?
The theme of the post was inspired by the news about the next Russian plane crash, this time in Kazan .
No, I will not speculate on the topic of who is to blame - not in this particular case, or in the aviation industry as a whole. What I really want is to understand why now, in 2013, the human factor can still be the cause of the plane crash, and also find the answer to the question: why do modern pilots need people?
Immediately, I’ll clarify that I’m not a pilot and I don’t have anything to do with aviation, so everything that is written below has a pronounced sofa character. Nevertheless, in my opinion, the post raises certain fundamental questions from the field of automation of the control of aviation technology - and we will all win if these questions have any answers.
So why don't airplanes fly by themselves?
Let me explain why such a question seems reasonable to me.
The fact is that for several years now there has been a project to automate the control of another type of vehicle, which previously was trusted only to people - by car. Various companies, including automobile manufacturers and universities , are engaged in this task, but Google’s project, where computer-controlled cars have already wound thousands of kilometers along California and Arizona’s roads, got into a single accident, and it was on the driver’s conscience most of all. man
')
But cars are by their nature completely unadapted for computer control. Of the sensors that can provide the computer with important information for decision-making, in cars there is only a speedometer and tachometer. Plus newfangled gadgets of recent years - parking sensors and cruise control. But orientation in space, adaptation to the current road conditions and current driving restrictions - all this was always left at the mercy of the driver, in whose assistance there were two main technical means - cuts in the body, called windows, and rear-view mirrors.
Especially titanic is the work done by the guys from Google, who supplied the car with a bunch of irrelevant sensors and developed a software part that allows them to read traffic lights, signs and markings, distinguish passing and oncoming cars, pedestrians and even animals crossing the road, as well as use this to effectively and safely reach from point A specified by the user to point B. indicated by the user.
Airplanes, if we take large passenger liners, which, of course, was the crashed Boeing 737-500 , are equipped with just a lot of sensors that transmit information about the speed and direction of flight, the height of the car above the ground, the angles of pitch, roll and yaw, and other-other-other . Moreover, airplanes are orders of magnitude smaller than cars, and each of them is always in the area of ​​responsibility of some air traffic controller who, if necessary, will provide clarifying information, such as weather conditions for the course or approaching aircraft, and suggest the necessary adjustments .
An important point is the preparation of the pilots themselves. If the driver of the car can learn traffic rules independently, practice driving in an empty parking lot and end up on the right with a minimum of problems, then civil aviation pilots will train on special simulators for months, and each type of aircraft requires individual training. Each action of the pilot is regulated by the instruction, and it is precisely the compliance of the crew’s actions to the instructions in a given emergency situation that determines whether the proverbial “human factor” will be called the cause of the incident.
Accordingly, the main question is:
Already in the process of writing this post, I found on Habré a note on a strikingly similar topic, but with the opposite message: " Can there be too much automation in the plane ?" Remarkable is the fact that it was published less than two days before the tragedy in Kazan.
In its article, its author, curiousGeorge , examines an example with a specific system of “smart” aircraft control and one specific problem of this system, which could cost the lives of the crew and passengers of one of the aircraft under its control, if not for the competent actions of the pilots. As a result of the analysis of the problem mentioned in the article, it is reported that the manufacturer decided to replace the control unit that caused the abnormal situation and seriously rework the control system software.
But what is interesting. The actions of the pilots of the ill-fated aircraft, which was almost injured due to lazy automatics, the author - judging by his words, the pilot himself - evaluates according to his own experience and qualifications. That is, the pilots did not give birth to the miracle of air piloting - they simply followed the instructions , which for some reason did not want to follow the kooky machine. Which once again confirms my guess:
And if so, what prevents to lay the same algorithm in the control computer?
In the comments to the post mentioned above there are concerns about the reliability of computers and the consequences of their failure. However, this fear seems farfetched to me. Everywhere automation replaces people or at least gives them the role of passive observers in most cases. Nuclear power plants, spacecraft, military equipment - all this is now extremely automated. And emergencies such as the blast at Chernobyl or the wreckage of the Costa Concordia only confirm the fact that a person is a much more likely source of risk than a car.
Well, if not a control system fails, but some source of data for it, the presence of a live pilot will not provide any additional reliability. A vivid example is the crash of the Airbus A330 in 2009, the cause of which was eventually recognized as icing of the speed sensors, after which the autopilot was turned off and the crew could not correct the problem with their inconsistent actions.
We live in a world where information technologies are closely intertwined with human life and machines sometimes provide a better result than people themselves. However, in aviation, there is a paradoxical situation where, with all the prerequisites for the full automation of the flight control process, the aircraft continue to fall, and people continue to die due to imperfect control systems and crew actions.
Why planes are still not completed with autopilot, able to completely replace the pilot in all situations, including emergency situations - the question is literally a million dollars. I propose by joint efforts to try to find an answer to it.
No, I will not speculate on the topic of who is to blame - not in this particular case, or in the aviation industry as a whole. What I really want is to understand why now, in 2013, the human factor can still be the cause of the plane crash, and also find the answer to the question: why do modern pilots need people?
Immediately, I’ll clarify that I’m not a pilot and I don’t have anything to do with aviation, so everything that is written below has a pronounced sofa character. Nevertheless, in my opinion, the post raises certain fundamental questions from the field of automation of the control of aviation technology - and we will all win if these questions have any answers.
The future is here
So why don't airplanes fly by themselves?
Let me explain why such a question seems reasonable to me.
The fact is that for several years now there has been a project to automate the control of another type of vehicle, which previously was trusted only to people - by car. Various companies, including automobile manufacturers and universities , are engaged in this task, but Google’s project, where computer-controlled cars have already wound thousands of kilometers along California and Arizona’s roads, got into a single accident, and it was on the driver’s conscience most of all. man
')
But cars are by their nature completely unadapted for computer control. Of the sensors that can provide the computer with important information for decision-making, in cars there is only a speedometer and tachometer. Plus newfangled gadgets of recent years - parking sensors and cruise control. But orientation in space, adaptation to the current road conditions and current driving restrictions - all this was always left at the mercy of the driver, in whose assistance there were two main technical means - cuts in the body, called windows, and rear-view mirrors.
Especially titanic is the work done by the guys from Google, who supplied the car with a bunch of irrelevant sensors and developed a software part that allows them to read traffic lights, signs and markings, distinguish passing and oncoming cars, pedestrians and even animals crossing the road, as well as use this to effectively and safely reach from point A specified by the user to point B. indicated by the user.
What about airplanes?
Airplanes, if we take large passenger liners, which, of course, was the crashed Boeing 737-500 , are equipped with just a lot of sensors that transmit information about the speed and direction of flight, the height of the car above the ground, the angles of pitch, roll and yaw, and other-other-other . Moreover, airplanes are orders of magnitude smaller than cars, and each of them is always in the area of ​​responsibility of some air traffic controller who, if necessary, will provide clarifying information, such as weather conditions for the course or approaching aircraft, and suggest the necessary adjustments .
An important point is the preparation of the pilots themselves. If the driver of the car can learn traffic rules independently, practice driving in an empty parking lot and end up on the right with a minimum of problems, then civil aviation pilots will train on special simulators for months, and each type of aircraft requires individual training. Each action of the pilot is regulated by the instruction, and it is precisely the compliance of the crew’s actions to the instructions in a given emergency situation that determines whether the proverbial “human factor” will be called the cause of the incident.
Accordingly, the main question is:
What exactly causes prevent the use in aviation of automated control systems that fully control the behavior of the aircraft at all stages of motion?
Underwater rocks
Already in the process of writing this post, I found on Habré a note on a strikingly similar topic, but with the opposite message: " Can there be too much automation in the plane ?" Remarkable is the fact that it was published less than two days before the tragedy in Kazan.
In its article, its author, curiousGeorge , examines an example with a specific system of “smart” aircraft control and one specific problem of this system, which could cost the lives of the crew and passengers of one of the aircraft under its control, if not for the competent actions of the pilots. As a result of the analysis of the problem mentioned in the article, it is reported that the manufacturer decided to replace the control unit that caused the abnormal situation and seriously rework the control system software.
But what is interesting. The actions of the pilots of the ill-fated aircraft, which was almost injured due to lazy automatics, the author - judging by his words, the pilot himself - evaluates according to his own experience and qualifications. That is, the pilots did not give birth to the miracle of air piloting - they simply followed the instructions , which for some reason did not want to follow the kooky machine. Which once again confirms my guess:
In any situation, the pilot has a previously known algorithm that he must follow in order to minimize the negative consequences for the crew, passengers and cars.
And if so, what prevents to lay the same algorithm in the control computer?
In the comments to the post mentioned above there are concerns about the reliability of computers and the consequences of their failure. However, this fear seems farfetched to me. Everywhere automation replaces people or at least gives them the role of passive observers in most cases. Nuclear power plants, spacecraft, military equipment - all this is now extremely automated. And emergencies such as the blast at Chernobyl or the wreckage of the Costa Concordia only confirm the fact that a person is a much more likely source of risk than a car.
Well, if not a control system fails, but some source of data for it, the presence of a live pilot will not provide any additional reliability. A vivid example is the crash of the Airbus A330 in 2009, the cause of which was eventually recognized as icing of the speed sensors, after which the autopilot was turned off and the crew could not correct the problem with their inconsistent actions.
Total
We live in a world where information technologies are closely intertwined with human life and machines sometimes provide a better result than people themselves. However, in aviation, there is a paradoxical situation where, with all the prerequisites for the full automation of the flight control process, the aircraft continue to fall, and people continue to die due to imperfect control systems and crew actions.
Why planes are still not completed with autopilot, able to completely replace the pilot in all situations, including emergency situations - the question is literally a million dollars. I propose by joint efforts to try to find an answer to it.
Source: https://habr.com/ru/post/202574/
All Articles