Source | Pinwancool
The author | ByArsT
"Obviously there is no car on the road, the navigation has to tell me that the road is congested, like a ghost, if in the middle of the night I dare not go."
I believe that many friends have encountered the above situation, obviously there are only a few cars on the road, the navigation map must show the current congested road section, so that many friends who do not know the specific situation have chosen to bypass the road section.
It happened to be during the Spring Festival, everyone began to leave for their hometown, and some friends around me encountered such a situation and questioned about this matter.
I asked a few colleagues around me who often drive, and indeed, almost all of the mainstream navigation software has such a situation, and if someone travels in the middle of the night, it will be really scary to encounter such a situation.
So what exactly causes the congestion information of mobile phone navigation to be "difficult to distinguish between true and false"?
Don't worry, if we want to figure this out, we have to start with the principle of navigation software "traffic information".
Collect "User Data"
Judge the road conditions
Navigation software wants to know the details of the road conditions, and intelligent analysis of whether the road section is congested, in fact, it is not difficult, in principle is very simple, the most important thing is the "user data" link.
In the era of big data, each of us in the open App to obtain convenience at the same time, but also to carry out "equivalent substitution", pay the corresponding information to free use, then at the navigation level, we pay is our driving data.
The driving data referred to here is the driving speed and driving route of the vehicle, through our mobile phone GPS information, the system can obtain the driving speed, location, direction and other information, and convert this information into traffic information, so that the navigation system can accurately determine whether the route is congested through big data.
If the GPS location information stays on a certain line for too long, then the system will suspect whether there is traffic jam or accident in this location. If the system is difficult to confirm, it will even pop up a notification asking if you are in a traffic jam or a traffic accident, so let us choose yes or no.
Of course, big data is not based on "individual data" to make overall judgments, it will also integrate all the user information on this road. For example, if ten signals appear on this road, and ten signals are in a slow state, the system will basically determine that the road is stuck in traffic, so that other users can avoid this congested route.
Some friends may feel that the data obtained in this way is not accurate, but in fact, there are already enough samples to provide data for the navigation system, and this is also the "bottom line" of these navigation software.
The current number of navigation software users is very large, only in April 2021, the average daily active life of the AutoNavi single platform has exceeded 100 million, and the current head navigation software is not AutoNavi. In 2021, AutoNavi's market share will be 33%, Baidu Map's will be 32.7%, and the two together will account for more than 60% of users.
In addition to navigation software, most of the car navigation is also with these navigation platforms to cooperate to obtain map information, at the same time will also upload your driving information to the server, so in the amount of data, these platforms have enough "bottom".
These platforms also have other channels for verifying that user data is accurate. For example, arbor navigation that has emerged in the past two years, while navigating for us, will also secretly look at our road condition information and whether it is consistent with the server information to further ensure the authenticity of the information.
Three kinds of data aid authenticity
But the "congestion information" is still inaccurate
As we just said, user big data is the main reason, so is there any other data to "assist" the authenticity of the information?
The answer is yes, and there is more than one, these data and user big data are different, these data are obtained from public channels, the overall summary is roughly divided into three categories.
The first category, that is, the user data we just said, as the fundamental source of the navigation platform data, user data can provide more comprehensive road condition information, such as some small roads, unfamiliar roads, or roads without supervision, user data can give a good answer.
The second category is the GPS information from the operating vehicle, such as the GPS information on the bus and taxi, which can reflect the road situation more realistically, and because the bus travels on a fixed route, after a long period of information verification, there is no need to worry about the accuracy of the road data, and even the road information situation in the next few days can be pushed out.
The third category is to cooperate with the traffic police department to obtain real-time road condition monitoring information, read the road license plate information through the identification camera on the road, which is a more authoritative official data, but this kind of data also has a more obvious defect, that is, there is a certain blind spot, many road sections do not have a camera that can identify information, and it can only be said that there is such information as road construction or major accidents in a certain section of the road, which can be accurately identified.
The integration of these three types of information constitutes the information network of the current navigation platform, and the nearby road conditions are completely analyzed.
But is the information integrated in this way accurate? Let's return to the question at the beginning of the article, what is the reason for the congestion information of mobile phone navigation "difficult to distinguish between true and false"?
Success is also Xiao He, and defeat is also Xiao He. Because the navigation platform has many sources of information, so that its information has a very high degree of authenticity, but intelligent algorithms can only carry out some shallow recognition, it is precisely so that it is difficult to understand some human behavior.
Sometimes there are "supernatural events" that occur, it is likely that several vehicles parked on the side of the road have opened the navigation, so that the system mistakenly believes that the road is seriously congested, so that many drivers who need to pass through this road have made "misjudgments" and detoured a little more in vain. When a user passes through this section at normal speed, the system finds an error and adjusts the speed to a smooth section of the road.
This is probably the true face of the "supernatural event".
Figure | "Live Nanyang" news report
A few days ago, there was a news that exploited this system vulnerability for profit, according to the Nanyang Daily "Live Nanyang" report, some Shanghai online ride-hailing drivers will be a large number of mobile phones together, all turn on the mobile phone's driving navigation function, and then placed on the ground without moving. In this way, the navigation system relies on big data to determine that there are many vehicles driving slowly or stopped in this section of the road, so as to automatically mark the road section red and set it as a congested road section.
Figure | Crack Google Maps project
Other unsuspecting drivers see congested sections on the map and make decisions not to take orders from around them. In this way, the counterfeit driver will get the order smoothly.
In 2020, foreign netizens also used this system "blind spot" to do a performance art, using a red cart to install 99 smart phones, turning on all the smart phones for Google Maps navigation, and pulling the cart forward slowly. On the map, the line becomes more and more "red" and more congested, yet in reality the streets are empty.
In the past two years, the navigation platform has had a good accuracy, especially in first- and second-tier cities. But even so, we still can't give the conclusion of which platform is more "easy to use", in the end, it is because of the algorithm, but it can only say a certain degree of intelligence.
Although these software now collect enough information to support the authenticity of the data, but the algorithm can not understand the full behavior of human beings, I hope that manufacturers can find a better direction for improvement, so that people who are in a hurry do not have to be wrongly guided.
This article is reproduced from Pinwancool (ID: pinwancool) with permission, if you need to reprint it for a second time, please contact the original author.