Translated by
2017/10/24 13:01:11

Auto pilot Unmanned vehicle

The automobile industry undergoes significant transformation: the largest producers of machines together with IT and a telecom developers go to creation of vehicles with a possibility of completely autonomous driving. The trend is already obvious – in the future pilotless transport will become the mass phenomenon, but on the way to an era of completely autonomous cars still it is necessary to solve the mass of problems. .

Content

The unmanned vehicle — the vehicle equipped with an automatic control system which can move without participation of the person. An auto pilot — the device or a hardware and software system conducting the vehicle on the trajectory defined, set to it. Most often auto pilots are applied to control of aircraft (because flight most often happens in the space which is not containing a large number of obstacles) and also to control of the vehicles moving on railway lines. The modern auto pilot allows to automate all stages of flight or movement and other vehicle.

General principles of work of the unmanned vehicle

The general principles of work at all unmanned vehicles are approximately identical. We suggest to get acquainted with them on the example of the Toyota Prius car in the version of Google.

The principle of work of the Toyota Prius unmanned vehicle in the version of Google. Secret-seo.ru
The principle of work of the Toyota Prius unmanned vehicle in the version of Google. Secret-seo.ru

Later developers come up with the idea to use high-precision cards together with the specified equipment. Autonomous movement only using sensors requires permanent scanning of the neighborhood and as result, huge computing powers. High-precision cards allow the car to move even on the roads which do not have a special marking, and sensors are supposed to be used only for timely reaction of the car to changes of situations on roads (crossing of the road by pedestrians, overtakings, etc.).

Technologies of the unmanned vehicle belong to the class of solutions of artificial intelligence. For more details see. Artificial intelligence (AI, Artificial intelligence, AI)

Standards for unmanned vehicles

Main article: Standards for unmanned vehicles

In a mra there is an active development of ITS of new generation with a big range of opportunities, such organizations as ETSI, IEEE, 3GPP and others are engaged in their standardization. The modern ITS systems solve such problems as admission control, management and payment by parkings, providing information on the movement and payment of the parking, management of cargo transportation, traffic observation, etc.

One of the main applications of ITS is the help to the driver of the vehicle. Due to cooperative awareness transport means the notification about dangers, the indicator of slowly going machines, warning of collision at the intersection can receive, the indicator about approach of the motorcycle, etc.

Notifications about such situations as breakdown of electric lighting, the incorrect road, the stationary machine (accident or breakdown of the vehicle), carrying out a roadwork, risk of collision, the notification about a status of traffic and the notification about change of a signal will be available to the driver. The decentralized databases will provide information on dangerous zones, rainfall, couplings on roads, visibility, wind, etc.

Use of ITS in unmanned vehicles will become the next step. External cameras and the radar equipment, according to the report of NIIR will be basic component of UAVs. But information exchange between cars on means of V2V-systems together with receiving by vehicles through the V2I-systems of information on a situation on roads and relevant digital maps of roads will allow to provide safe and effective traffic of UAVs.

Types of ITS: V2V and V2I

The first type of systems - "the vehicle-the vehicle" (vehicle-to-vehicle, V2V) - provide safe driving due to communication between cars at intersections with poor visibility. A V2V-system can warn drivers about danger of head-on collision, side crash, back collision, notify on fault of the vehicle, provide the road and regulatory information

For example, two machines invisible each other at the intersection or on turn, through a V2V-system can exchange with each other coordinates and values of speeds for collision avoidance. The same way the car approaching the end of a stopper will obtain information with coordinates and speeds of the next vehicles.

The second type of systems safe the movement - "roadside infrastructure - the vehicle" (vehicle-to-infrastracture, V2I) - provide information transfer (a signal and the regulatory information, etc.) from the roadside equipment to the car through means of a radio communication. For example, roadside sensors at the intersection will detect machines which are going to cross the intersection or to turn, and will transfer information to other approaching machines on means of V2I-systems.

V2X technology: use of Wi-fi and cellular networks

Main article Vehicle-to-Everytning (V2X)

Degrees of autonomy of cars

On classification of SAE International of the systems of the help to the driver or ADAS (Advanced Driver Assistance System) there are six classes of autonomy from level 0 — completely manual control with a possibility of warning of dangerous situations on the road, to 5 — completely unmanned vehicle. The ADAS 1 level provides work of more advanced alerting system about danger of collision with cars, pedestrians and also about intersection of the line of a marking, identification of road signs, etc. and also intervention in a management system. The second level is more active help to the driver (taxing, braking, deduction in a band, etc.). The third level — the autonomous movement on the set road sections which demands from the driver only of partial supervising control. The fourth ADAS level is the autonomous movement of the car in certain modes at which the person cannot influence management even in critical situations any more. The fifth level — complete autonomy of the vehicle when the driver only sets a terminal point of a route, and all process of movement completely lays down on artificial intelligence of an auto pilot.

Advantages and shortcomings

Advantages

  • transportation of goods in dangerous zones, during natural and technogenic catastrophes or military operations.
  • reduction in cost of transportation of loads and people due to economy on the salary of drivers.
  • more economic consumption of fuel and use of roads due to centralized operation by traffic flow.
  • saving of time which is nowadays spent for management of the CU allows to be engaged in more important issues or to have a rest.
  • people with the weakened sight have an opportunity to move independently on the car.
  • minimization of road accident, human victims.
  • capacity growth of roads due to narrowing of width of road bands.

Shortcomings

  • Responsibility for causing [1]
  • Loss of a possibility of independent driving of the car.[2]
  • Reliability of the software.[3]
  • Lack of experience of driving at drivers in critical situation.[4]
  • Loss of jobs by people, whose work is connected with driving of vehicles.[5][6]
  • Loss of privacy.[7]
  • Mining of unmanned [8]
  • The ethical question of the most acceptable number of the victims similar to a trolley problem facing the car computer at inevitable [9] [10]

Some systems rely on the infrastructure systems (for example, built in the road or near it), but more advanced technologies allow to simulate presence of the person at the level of decision making about taxing and speed, thanks to a set of cameras, sensors, radars and systems of satellite navigation.

Calls on the way to the UAV

At the beginning of 2017 the correspondent of TAdviser visited the Mobile congress in Barcelona and examined development of technologies of unmanned vehicles.

Communications

On the way to unmanned vehicles still it is necessary to solve many technology and legal problems. Developers agree in opinion that one of key – to provide cars with opportunities of high-speed network connection. Networks of fifth generation are considered as the driver of technologies of autonomous driving: they will allow the car to obtain most quickly information and to interact with other cars and the infrastructure surrounding it.

The minimum transfer lags of information which are expected in a 5G are critical for unmanned vehicles at their mass use. High-speed communication will allow to accept and transfer instantly data from one car to another. Information on changes in the movement of one car, for example, about braking, will allow to adjust operations of the machines surrounding it at once.

As of the beginning of 2017 of the standard of communication of a 5G does not exist yet. Regulators, the world telecom companies and equipment manufacturers are involved in its development. 3GPP (3rd Generation Partnership Project) — the organization approving international standards of cellular communication – is going to complete completely the work on testing and standardization of technologies of wireless communication of fifth generation in 2020.

In February, 2017 the International Telecommunication Union published the first version of the working draft copy of the specification describing 5G network. The draft document sets a level of the expected performance of new IMT-2020 standard: it is supposed that the average speed of downloading in a 5G networks for users will be 100 megabits per second, and loadings — 50 Mbps. At the same time waiting time will not exceed 4 ms (for 4G this LTE the value is about 20 ms).

5G networks should accelerate mass implementation of technologies of pilotless driving
5G networks should accelerate mass implementation of technologies of pilotless driving

The special systems allowing the car to exchange data with other objects are also developed for communication with surrounding objects. The vehicle-to-everything (V2X) technology for wireless communication allows the car to receive warnings of road conditions and the approaching cars long before they appear in its viewing field. For this purpose and surrounding infrastructure should be "smart". For example, traffic lights, road marking, road signs.

Development of the interface for the V2X systems which will be able to work with networks of new generation is conducted, for example, by Qualcomm. The company tells that are going to hold until the end of the year testing of a prototype as a part of end devices in Germany in partnership with a number of the companies, among which Ericsson, Audi.

In March, 2017 the head of development of mobile online services of Volkswagennikolay Nikolai Reimer noted that one of key tasks – to provide machines with communication opportunities. The company considers this question so important that it even purchased about three years ago the European research and development center of BlackBerry with a team approximately of 200 engineers.

Based on this division of Volkswagen develops Competence Center according to solutions of communication for the cars now. It provides development of technologies which will be able to be applied in the connected cars in the future. Among them – communication control units. Volkswagen expects to provide itself with them in the future. Nicolai Rhymer considers that the company should invest more in these developments.

Eye of Omniscience

The autonomous car should know with an accuracy of centimeters, where exactly it is also that is further on the road out of a zone of the current physical visibility. The cartographic company Here (Nokia belonged earlier) note that cards of high accuracy – a fundamental element in addition to sensors and cameras in order that the unmanned vehicle could be guided in the situation surrounding it.

Cards should reflect also location of the car and to allow it to know what is further, behind turn what cameras and sensors cannot provide. Then the car will be able to build not the reactive, but pro-active strategy of driving, Alex Mangan, the head of product marketing of Here says.

For testing of the Google unmanned vehicles previously itself, for example, builds the detailed 3D maps on pilot routes considering even small features of roads. For data collection on the basis of which the card will be under construction the staff of the company previously specially goes on roads. In a case with test routes it is a feasible task, however, when it is required to create cards for roads millions kilometers long, it looks difficult implementable. Especially taking into account that once the created cards need to be supported and updated – the picture on roads can change very often.

The autonomous car should know what occurs not only in a visibility range, but also behind turn
The autonomous car should know what occurs not only in a visibility range, but also behind turn

Cooperation with car makers can simplify creation of accurate maps for cars: their machines equipped with sensors and radars can "share" information obtained from roads with developers of cartographic services. At the expense of it cards could be updated literally in real time.

In February, 2017 the decision maker for Mobileye and BMW unmanned vehicles announced similar cooperation. Its purpose is collecting of geospatial data for self-governed machines. The BMW 2018 cars of model year will be equipped with cameras and software of MobilEye for collection of information, necessary for updating of digital maps of high resolution.

For the purpose of the accelerated creation and updating of the BMW and Mobileye cards will transfer the data generated within partnership, Here company. Alex Mangan from Here believes that the industry should integrate around the idea of data exchange – it can accelerate distribution of technology of pilotless driving. In addition to BMW, the company is going to agree also with other producers of cars about similar data transmission, including Audi and Mercedes.

Recognition of road signs and marking

As researchers managed to detect at the beginning of August from the university of Washington, the systems of machine vision applied in machines with an auto pilot to recognition of road signs it is easy to confuse: for this purpose it is enough to place definitely on signs small stickers.[11]

During the experiment researchers pasted several black and white stickers on one of signs of Stop, on another placed additional texts on top and from below from Stop text, and made the third sign more faded. At the same time it is emphasized that in all cases road signs remained quite known and were read well.

Photo: 24gadget.ru
Photo: 24gadget.ru

Nevertheless, the autopilot system in the majority of attempts glitched: the above described manipulations with signs of Stop led to the fact that instead of them the auto pilot "saw" the restrictive sign of speed.

Photo: 24gadget.ru
Photo: 24gadget.ru

Results of an experiment suggested researchers an idea that malefactors can independently do similar stickers to force a computer system of the car to recognize incorrectly the sign of traffic.

As a method of fight against the detected vulnerability researchers suggest to implement the algorithms which are in addition analyzing a context in which the sign met in the system of an auto pilot. In particular, algorithms will help a system to define that the sign is located in the inadequate place (for example, Stop — on the high-speed motorway or restriction of speed in 100 km/h — on the city street) that will help to avoid an emergency situation.

As noted, not the system of some specific car maker, but an auto pilot operation algorithm, standard for all producers, was subjected to test. Results of a research, according to authors, show degree of vulnerability of automatic equipment. The distortions used on road signs modelled the typical striking factors of the urban environment: acts of street vandalism, I spoil sign coverings because of weather conditions and so on.

During the experiment scientists used several road signs with different types of texts, stickers and graffiti. According to researchers, in 100% of cases, cars recognized the sign "Feet" with Love\Hate texts as the sign "Restriction of Speed 45", the second and third signs were also recognized as "Restriction of speed 45", but only in 67% of cases. As for the fourth sign, its system of machine learning classified as the sign "Feet" instead of "The right turn" in 100% of cases.

Security

In October, 2017, speaking at the World forum of knowledge in Seoul, South Korea, the chief executive officer of Mobileye and the senior vice president of Intel, professor Amnon Shashua offered the automotive industry the method allowing to confirm safety of unmanned vehicles. Its solution published in the scientific article and provided in a summary of this work for inhabitants offers a mathematical formula, using which it is possible to confirm that this or that unmanned vehicle works with observance of regulations of responsibility and cannot serve as a cause of accident, the blame for which could be laid on this car.

The Responsibility Sensitive Safety model provided by scientists provides the specific, giving-in to measurement parameters characterizing human ideas of responsibility and care and defines so-called Safe State, supporting which the unmanned vehicle cannot serve as a cause of accident, regardless of what maneuvers or actions make other vehicles.

In the performance Amnon Shashua called industry members and those who develop strategies, "together to work on creation of standards which would allow to set unambiguously responsible" at inevitable accidents with participation of the cars managed by drivers and unmanned vehicles. He explained that all modern rules and regulations are based on that idea that the driver therefore for regulation of unmanned vehicles it is necessary to enter new parameters into rules drives the car.

"The key moment is now an opportunity to set responsible for accident. Even the best drivers in the world get into the accidents, and unmanned vehicles will not be able to avoid this fate because of actions of other participants of traffic too. But the probability that the responsible and careful driver will have an accident on own fault is very small, especially if the panoramic overview is available to the driver, and it has flash-like reaction – as at the unmanned vehicle", – Amnon Shashua explains. The offered RSS model allows to formalize work of unmanned vehicles therefore "UAVs" will always work only within that model which is considered "safe", proceeding from accurate determinations of fault and which is approved by industry members and regulatory instances.

Cyberthreats – one of calls for any attached devices, including cars. The head of telecommunication company SoftBank Masayes Dream (Masayoshi Son) at the end of February, 2017 provided data that the number of cyber attacks to objects with Internet connection grew four times in 2016 in comparison with 2015.

In a case with cars it is the special reason for concern as as a result of actions of malefactors I can suffer people. Theoretically, the hacker can hack network, stop data transmission, switch off brakes or just stop the machine.

In the middle of 2015, for example, specialists in a computer security of the Center of advanced technologies of Uber detected vulnerability in the software of the Jeep car thanks to which could provide remote access to some systems of the car: to the conditioner, window wipers, audio system and brakes.

«
Cyber-incidents are a problem for any car maker in the world. It is a question of public security, - the CEO of General Motors noted before Mary Barra.
»

The Argus company specializing in means of cyber defense for cars considers, any uniform product cannot be suitable for these purposes: the different solutions intended for different parts of the connected car should be integrated among themselves that security blanket was provided.

Car makers and decision makers for cars invest in development of this direction of cyber security. A number of car makers, including Tesla, Fiat Chrysler and General Motors, created special programs of encouragement of persons which will announce gaps in security of the systems of the machines.

Responding to the market requests, appears more and more companies developing specialized solutions for cars. Such direction appeared, for example, and at Kaspersky Lab. In 2016 the company reported that it conducts development of the protected safe operating system which, in particular, can be used for cars.

In 2016 the decision maker for the Harman cars purchased the Israeli technology of cyber security TowerSec for protected from software cracking. The same year investments for development of technologies of automobile cyber defense were received by other Israeli startup - Karamba Security.

Whose life is more important: driver or pedestrian?

In addition to technology calls to go to mass use of UAVs it is necessary to resolve also the "moral" issues connected with decision making by an auto pilot. For example, whether it should be designed so that to protect the driver's life at any cost even if in emergency situation for this purpose it is necessary to ram crowd of pedestrians?

Rules for unmanned vehicles in Germany

The federal Ministry of Transport and Digital Infrastructure of Germany announced intention to enact road rules for unmanned vehicles, their producers and owners. According to in the summer of 2017 of The Register, the document will oblige developers of unmanned vehicles to program their auto pilot so that in any unexpected situation on the road it put human life above life of animals or safety of private or state-owned property[12].

Now in one country of the world there are no uniform traffic regulations which would regulate technical requirements to unmanned vehicles and regulated their movement on public roads. Some countries allow movement of unmanned vehicles on the general roads, but in this case obtaining special permission is required. At the same time driving the unmanned vehicle always there has to be a driver ready to intercept at an auto pilot management in case of any emergency situation.

Rules for unmanned vehicles in Germany were developed by the advisory board of the Federal Ministry of Transport and Digital Infrastructure of Germany into which 14 scientists and lawyers enter. In total 20 requirements to pilotless machines, their producers and drivers were included in the list. So, the requirement to the value of human life means that the auto pilot in any emergency situation should drive thus the car to save life of people.

In case of a dual emergency situation, the auto pilot should not make the choice whose life it is necessary to save — the driver or the pedestrian, the elderly person or the child. The auto pilot will have to do everything possible to save life of all participants of accident. All unmanned vehicles registered in Germany should have "a black box" from which records can be used after accident to find out on whom responsibility for it — on the driver or on an automatic pilot lies.

At the same time in all cases of accident with participation of the unmanned vehicle "the guilt presumption" will work, i.e. in accident the driver will be always considered as a guilty person, so far data of "the black sensor" or other results of investigation of incident will not prove the return. Also included the exclusive right of drivers to the choice of information which his producers will be able to receive from the unmanned vehicle in number of rules. It is about location, speed, driver's data and a set of other information which can be used, for example, for advertizing targeting.

Moral dilemma

The psychologist to School of economy in Toulouse Jean-Francois Bonnefon and his colleagues tell[13], people in general support the idea that in critical situation the car should crash into a wall or still somehow offer the driver to save bigger number of pedestrians. At the same time the same people want to go in cars which protect the driver at any cost even if it will entail the death of pedestrians.

Such conflict puts producers of the computerized cars in a difficult situation, Bonefon notes. Between the car which is programmed for the benefit for the majority and which is programmed for self-defense of the passenger buyers in the majority will select the second.

There are many scenarios of emergency situations when the car has to make a choice whom to offer
There are many scenarios of emergency situations when the car has to make a choice whom to offer

Authors of the research about a social dilemma of autonomous cars published in the Science magazine in 2016 believe that in it areas are also other difficult moral questions. Autonomous cars should make decisions which effects cannot be predicted in advance in emergency situations. Whether it is admissible to program, for example, the machine on that it avoided collision with the motorcyclist, having crashed into a wall? The passenger of the car in this case has more chances to survive, than at the motorcyclist who will face the car.

«
Autonomous cars can make a revolution in the transport industry, but they put a social and moral dilemma which can slow down distribution of this technology, - Iyad Rahwan, the scientist from the University of California, one of authors of this research considers.
»

The psychologist Kurt Gray from the University of North Carolina is confident in Chapel-Hil that it is possible to reach the working compromises. If unmanned vehicles also are programmed to protect the passenger in emergency situations, the number of road incidents anyway will decrease. Except for exceptional cases when such cars can constitute danger to passengers, they anyway will not exceed speed, will not take alcohol or to type text messages on the run why society, eventually, will win.

The moral of artificial intelligence — one of the most discussed issues connected with approach of an era of robots. In 2016 the Massachusetts Institute of Technology (MIT) in the USA developed the special test which helps to understand better what moral dilemmas developers of artificial intelligence face, and also to deal with the moral guidelines[14].

Test very simple. In it it is necessary to deliver himself to the place of artificial intelligence of the self-governed machine and to select who in road accident can be offered — pedestrians at the intersection or passengers in the car. Sometimes it is necessary to select meanwhile which of pedestrians it is necessary to press and whom — to save.

In total in the test of 13 questions. The number of passengers and pedestrians in some tasks a miscellaneous, in some — identical. Besides, their age, sex, a social status differs. In some questions there are pets — they are equated to other passengers and pedestrians.

For an example we will take the following task: in the machine the woman and two children (the boy and the girl) sit, and on the crosswalk there are a woman and two old women. It is necessary to select which of them to artificial intelligence to save and whom — to sacrifice.

At the end of the test the user is told whom he sacrifices most often and as other people who passed a test answered.

On the website of MIT it is also possible to think up own moral task on the basis of questions of the test and to look what questions were raised by other users.

Legislation

In addition to technology calls, to go to mass use of autonomous cars it is necessary to resolve a set of issues at the level of legislative regulation. The regulating documents defining the basic technology and legal conceptions in this sphere, regulation of opportunities of use of such technologies in general, responsibility in case of incidents with unmanned vehicles, etc. are necessary.

In this or that type regulating documents in this area are already submitted or are developed in some countries. Especially moved ahead the USA here. Nevada in 2011 became the first state in the country which began regulation of use of autonomous vehicles on roads and the questions connected with their insurance, security and testing.

Conditions of movement of unmanned vehicles of different degree of freedom are now legislatively fixed in different states of the USA. In 2015 the governor of Arizona, the USA, signed the order according to which unmanned vehicles in the state are registered on the same conditions, as normal machines. No additional requirements were imposed to independent machines. Besides, state statutes do not prohibit testing of unmanned vehicles on roads.

At the end of 2016 the governor of the State of Michigan signed a packet of laws which directly concerns the sphere of unmanned vehicles and actually legalize their private and commercial use. They allow sale of serially released unmanned vehicles which underwent certification, at the same time it is authorized to cars to come out to public roads without driver driving and to move in motorcades. Besides, in the territory of the state it is authorized to use the pilotless taxi now.

In Britain in 2016 began preparation of amendments in the legislation which should, first, allow to insure responsibility of pilotless machines, and secondly, to update the Road code (the code traffic regulations of Great Britain) taking into account development of autonomous vehicles.

News and models of producers

In the world

In Russia

Pilotless trucks

Main article: Pilotless trucks

Unmanned vehicles in Russia

Main article: Unmanned vehicles in Russia

Since 2015 the companies in Russia actively develop the technologies necessary for creation of unmanned vehicles.

World market

Main article: Unmanned vehicles (world market)

Creation of the full-fledged unmanned vehicle - one of the most fascinating calls for a technology thought of the beginning of the 21st century for the companies worldwide.

See Also

Robotics



Notes