Although smart machines should be self-contained and rely on nothing more than external GPS data (if they use GPS at all), the best solution would be to integrate them into a common data network. This network will appear soon, and unlike modern cloud services, it will not stop at the initial version, but will continue to improve over the next few years.
Presumably, this will be a high-level data transmission system. Information will be transmitted through broadcasting channels in the city, on the streets and at intersections. In addition, the signal from the machine to the network services will go through a two-point PPP line. The IT community also plans to establish direct communication between transport units (V2V) and between a transport unit and roadside infrastructure (V2I).
Since smart cars must consider everything from traffic signs to pedestrians, data transmission alone is not enough for the safe operation. Although there is definitely a plus from her.
About the author: Brad Templeton – A software engineer, an evangelist of robotic cars since 2007, worked on Google in his early years. Founder ClarinetHonorary Chairman Electronic Frontier Foundation and director Foresight institute, founder of the faculty in Singularity university.
Map and Database Updates
Data storage is so cheap that all the machines connected to the network are able to fully memorize the maps of the road system. Not only the position of the streets, but also completely everything that is on them: driveways, house numbers, lines, traffic lights, etc. A good smart machine is able to recognize all these things with its built-in sensors, but the data warehouse is insurance. So, if the sensor readings and the data are not consistent, then the stored information will help further assessment. If the database or car sensor notifies you that the “STOP” sign is approaching, the smart car makes a safe stop. In addition, the database can also be used to plan future movements.
If an error has occurred in the database, then either passengers or automatic systems of the machine will report it. The error will be quickly fixed, and the database restored in a short time. Of course, the car will not go just because the database reports the presence of the road. A smart machine will constantly confirm information with sensors. But perhaps it will be necessary to switch to manual control where the database erroneously reports that the building is not the road ahead.
Real-time traffic and traffic data translation
“Broadcast” will transmit the schedule of each traffic light, show the density of cars and the speed of movement of vehicles. So smart cars will be able to plan their trips and choose the best route. Also, broadcasting will not only transmit relevant information, but also predict the traffic situation of the future. The forecast will be calculated using algorithms and information received from vehicles located at the desired distance.
Information technology support includes actual broadcasting from traffic lights using the I2V protocol. But since expensive upgrades of intersections cannot be dispensed with, implementation will be slow. But sometimes traffic lights are already managed by central servers. In this case, using mobile networks, it will become possible to predict traffic switching for 1-2 seconds ahead. (Push-button traffic lights must be translated into I2V protocol.)
Travel time reservation
A smart car can interact with a central computer to book various services, including finding a parking space or calling a lift. Over time, the work of the traffic light will be regulated depending on the movement of smart cars. The four-way stop sign can turn into a virtual traffic light for smart cars. A car can safely cross such a traffic light if at that moment there is no freight transport or pedestrian at the intersection.
Some programmers have even prescribed algorithms that include traffic control at the intersection. Each smart car can reserve a time window during which it is going to move. Accordingly, the rest of the cars at this time will let her through.
Earlier, I talked about simulating such a system at the University of Texas at Austin. Check out the 6 lane version. I think people will be afraid of this system until they are convinced of its full reliability.
You can also book travel times in areas with restricted traffic. Because, what’s the point of getting stuck in traffic. And in fact, with proper regulation, not a single section of traffic will be overloaded. The best option is for each zone to determine the norm of the number of cars so that you can move around freely. Because there is nothing good either in flashing into a red light, or in forming a traffic jam. And elementary should be responsible for this distance sensors. And if too many cars are approaching one section, then some of them will be redirected to another road.
If the time requested for travel is unavailable, the smart car will simply slow down and then arrive at the deadline. If the desired time interval is already taken, then the start of the trip is delayed. More advanced measurements will allow passengers not to waste time and not to stand on the sidelines, but to wait for the car at home.
Signal to clear the road
Smart cars will be able to park anywhere, and even where it is forbidden to block the passage, for example, on the driveway. The transmitter from the roadway will send the necessary signal that for a while you need to clear the way. When you return home in a smart car and you need to park it in the driveway or put it in the garage, a minute before your arrival the car will signal the house, and it will signal to all the other cars on your way that you need to drive off. Smart transport instantly responds, and you drop into your parking lot without any hassle.
When you need to leave, the signal will come again. True, any advanced smart car will leave the roadway as soon as it notices that you need to drive.
This protocol will significantly increase the number of parking spaces in modern cities. In the absence of a busy traffic stream, smart cars will be able to park on a multi-lane road in two and three rows, leaving only one lane for traffic. So, in New York and other major cities there will be a sufficient number of parking spaces.
Smart cars will recognize passing vehicles and interact with them. They will be able to determine not only which car is on the way, smart or truck, but also which model it is. Such recognition will provide information on the characteristics of the vehicle, for example, its braking distance, reaction speed, maximum speed, maneuverability, etc.
In addition, they will receive a digital safety certificate of vehicle reliability. In addition to the reliability of the characteristics, the certificate will confirm that, if necessary, the car can adapt to the actions of another smart car.
For example, based on a behavioral model of a school of fish, when you see an obstacle, you need to collapse. But with heavy traffic, when there are other cars on either side of you, you will not be able to turn. Only if the neighboring car does not give you a signal that there is free space next to it, it will turn there and give you a place if you ask about it.
But, of course, any smart car in such a situation will give way. If a smart car sees a vehicle moving on it, it, like a person, should quickly orient and move to a safe place. If a certificate of reliability is obtained, then the smart machine will decide that this situation is caused by someone’s need to turn. In ordinary circumstances, a smart car is sure that it has a “way out” of any situation, for example, if a pedestrian runs out onto the roadway or an accident suddenly occurs ahead. When the movement is not very busy, there really is a way out, in the form of a place for maneuver. And with a regular stream, a skilled driver will always find a place to taxi.
Nevertheless, if you decide that you can rely on other vehicles, and they can quickly maneuver, then you can approach them closely, without fear of a collision and other obstacles. Of course, if you are not sure about another car, an accident may occur – but the culprit will be recorded.
As a rule, smart cars are certified. The basic certificate ensures that they are safe for use on the road. What, incidentally, the presence of a driver’s license can not assure. Private companies can conduct certification at a higher level, so you will know which cars you can rely on in an emergency.
Freight transport will not be as reliable as a car – it will need more space for maneuver. Here’s what it looks like today. Experimental smart cars and smart cars modified by the owner will not receive a certificate of reliability and, therefore, they will need to be kept at a longer distance.
Of course, all cars, reliable or not very good, freight or smart, must transmit data signals that they suddenly have to brake or turn. This is simply a more reliable version of the brake lights and turn signal, which smart cars will also be equipped with and will be able to decrypt them.
Mesh networks and communications between transport units (V2V)
Although, in general, the amount of information in Motornet for the entire network will not reach large sizes, if it does, then the machines will be able to connect to a network of cells in which large amounts of information will be transmitted via short high-speed circuits.
So you still cannot rely on this type of communication, important data will be transmitted to fixed cells. If the density of cars connected to the network of cells increases, then the mesh network can provide reliable communication and transmit the current situation. But this is still a long way off.
Current information on the situation on the road is very useful. Each car, driving along its route, captures bumps, potholes, puddles and ice. Information that the wheels are losing traction can be recorded and transmitted to other cars that follow.
Since GPS does not always detect such objects, optical sensors on the road and on the side of the road will allow cars to get more accurate information. The car will know in advance when it will encounter a glaciation of the road and prepare for it.
Overlapping signal transport
If vehicles communicate their location using a connected vehicle / ITS (a radio system known as DSRC, 802.11p, WAVE, etc.), this can be used to detect vehicles that are not visible to sensors or even visible to humans. With the mass introduction of such radio stations, it will be possible to prevent a small number of accidents. Attempts are being made to oblige all cars to install such radios. A decision is expected in 2013 and, if approved, in 2019 the radio stations will appear in new cars.
Of course, smart cars will have a system through which they can be called or hired, as well as send messages to their owners.
Typically, a smart car operates autonomously, following the instructions of its passengers or owner. In some cases, passengers or the owner may switch control to autopilot. For example, when entering a large parking lot, a smart car will allow the server in this section to decide where it will park and how to get there. Passengers got out and instructed her to listen to the parking server signal until they returned.
Convoys will also require transfer of control.
Many types of communications are useful for finding a parking space. Smart cars are able to make a request to the “parking space market” and using the calculation algorithm to find the nearest, most favorable condition for parking. (Of course, the algorithm will take into account the places in the nearest free parking and send the smart car there).
Parking costs for a smart car can be lower than for freight vehicles. In particular, you can include the following points in the parking rules and then make a discount:
- The smart car is heading to remote parking spaces that are not popular among people-driven vehicles. For example, places in the corner zone, around the perimeter or on the upper floors of the parking lot.
- Having reached the place and making sure that only other smart cars are parked around, the car gets close to them.
- When a locked smart machine needs to drive out, it will give a signal (by contacting the database or through direct WAN transmission, flashing lights or simply indicating forward movement). Other cars successively surrender and, having released the car, will stand back. The longer the car is parked, the more distant a parking place it should take.
- At the server’s signal, the smart machine should leave its place, even if its parking time has not yet expired. Otherwise, the parking fee will be increased.
The last point allows the server to sell smart machines cheaper seats. For example, the parking status may indicate that there are “no seats”, but when a new full-fledged client appears, the computer will simply ask smart cars to free up space. Cars will drive away, and the client will freely take the vacant seat. Thus, there will always be places for full-fledged customers, and smart cars will not prevent parking from making money on them.
A situation may arise that, during movement, a smart machine will require competent human intervention. It may happen that a passenger loses consciousness or gets into an accident and needs medical attention (an accident may very well be caused by another driver).
Complications can be associated with the closure of the road due to an accident or repair work (for which there was no information in the database), with unforeseen events along the way or natural disasters, for example, a flood or earthquake.
In such cases, the vehicle must act in accordance with its safety instructions. Next, the remote operator takes control and finds a way out of this situation. For this, the operator will need data from sensors, photos, audio and video files. Due to the delay in data transfer, he will not be able to fully control the machine, but he will be able to lay the route along which smart transport will follow using his automatic systems. In an emergency, the operator can slowly operate the machine, relying on the built-in obstacle recognition system. In case of urgent need, the operator can ask the car not prescribed actions by stepping over the algorithm. For example, move the box, after hitting it with a bumper and making sure that it is empty.
The above requires good network communication, which is not on every section of the route. Therefore, while autonomous means should avoid the terrain uncovered by such a network.
We are perhaps the most powerful competence center in Russia for the development of automotive electronics in Russia. Now we are actively growing and we have opened many vacancies (about 30, including in the regions), such as a software engineer, design engineer, lead development engineer (DSP programmer), etc.
We have many interesting challenges from automakers and concerns driving the industry. If you want to grow as a specialist and learn from the best, we will be glad to see you in our team. We are also ready to share expertise, the most important thing that happens in automotive. Ask us any questions, we will answer, we will discuss.
Read more useful articles:
- Cameras or lasers
- Autonomous cars on open source
- McKinsey: Rethinking Software and Electronics Architecture in Automotive
- Another OS war is already under the hood of cars
- Program code in the car
- In a modern car, there are more lines of code than …