CN104236566A - Map matching method based on intelligent mobile phone - Google Patents

Abstract

The invention discloses a map matching method based on a smart phone, which includes: installing an APP with a data collection function on the smart phone; fixing the smart phone in a vehicle, and opening the APP with a data collection function; Road, and manually mark the event to obtain the acceleration sensor data; obtain the corresponding relationship between the smartphone coordinate system and the vehicle coordinate system; correct the acquired acceleration sensor data; train and classify the marked and corrected acceleration sensor data to obtain the road Discriminant model: collect the measured road condition data, judge the road category according to the road discriminant model, and realize map matching in combination with road topology information. The present invention uses mobile phone sensors to detect curves and correct existing navigation systems to compensate the accuracy of civilian GPS systems and inaccuracies of map systems to a certain extent, provide more accurate navigation services, and make driving behaviors safer.

Description

Smartphone-based Map Matching Method

technical field

The invention relates to the technical field of digital maps and map matching, in particular to a smart phone-based map matching method.

Background technique

With the development of the economy and the increasingly developed road traffic, people's travel has become more convenient. The traditional map has long been a history, and the Global Positioning System (Global Positioning System, GPS) provides navigation assistance for people's travel today. The GPS system uses the transmission and correction of the position signal of the receiving end transmitted by multiple satellites to realize the position positioning of the receiving end. The constellation system composed of 24 satellites can almost cover the whole world. Now our common vehicle system is the combination of GPS system and geographic information system (Geographic Information System, GIS), using the map information of the road position in the electronic map to compensate the problem of low positioning accuracy of civilian GPS to a certain extent. With the improvement of wireless communication technology, the GPS receiver can be reduced to the size of a sensor and packaged in a mobile phone. Therefore, the mode of smart phone platform equipped with a navigation system is becoming more and more popular, and a large number of applications of advanced driver assistance systems have been spawned, which can provide more comfortable , Better efficient, more secure service. It is now possible to replace the function of traditional car maps with mobile phones.

However, in the case of complex road conditions, whether it is traditional car maps or mobile phone navigation, there are situations where positioning is inaccurate and some geographic information is missing, especially when vehicles pass through tunnels and curves. Therefore, this patent proposes a modified vehicle navigation and driving assistance positioning system based on a smart phone platform and using a mobile phone sensor to determine the position information of the vehicle's curve state.

The rapid development of smart phones, the global market research company (Gartner) data show that in the second quarter of 2013, the global sales of smart phones surpassed that of feature phones for the first time. At the same time, the survey in the second quarter of last year showed that the global market share of the open source mobile phone operating system Android (Android) system was 79%, forming a dominance situation. One of the reasons why people are so fond of smart phones is that compared with traditional mobile phones, smart phones have built-in many simple and convenient mobile phone applications (Application, APP), which can provide people with a more comfortable and convenient life experience. Many of these apps benefit from the many sensors integrated inside the smartphone. Under normal circumstances, the smartphone will integrate sensor devices such as temperature sensor, gravity sensor, distance sensor, electronic compass, light sensor, three-axis gyroscope, infrared sensor, etc., which provides a powerful and convenient platform for subsequent APP development. .

The GPS system originated in the United States in the late 1950s and was put into use in the 1960s. It was originally designed for military use, and then gradually opened up to civilian use, but its positioning accuracy is far lower than that of military use. The entire system consists of a constellation of 24 satellites, and the positioning range covers 98% of the world. Generally speaking, navigation satellites and ground receiving equipment are required to complete common navigation functions. The navigation satellites continuously send out the time and position of the satellites, and transmit them cyclically at intervals of 30 seconds; the ground receiver receives the messages sent by the navigation satellites and measures the The distance between the known position satellite and the receiver is corrected by integrating the data of multiple satellites. Generally, the correction process requires information from at least 4 satellites. The general car map is to use the GPS system for navigation. However, for open civil codes, due to satellite clock error, satellite ephemeris error, ionospheric delay error, tropospheric delay error, multipath effect error, receiver noise error, occlusion and other reasons, the accuracy can only reach about 20m ; In the case of higher map accuracy, the navigation accuracy will be improved accordingly. However, in curves, especially in dangerous road conditions such as winding mountain roads, the positioning accuracy of navigation is still not ideal, which requires the correction of technologies such as curve detection to improve the accuracy.

In the existing technology "Research on Map Matching Algorithm Based on Topology Structure", the positioning data of the vehicle and the electronic map data are registered and corrected by using the method of software correction to reduce the display error between the road information of the electronic map and the GPS positioning information . Calculate the weight of roads to be matched, adopt fuzzy matching strategy, remove disconnected roads, and finally use linear interpolation method for map matching. Through the test, the matching accuracy of the matching algorithm is not lower than 96%, and the single-point matching time is not more than 10ms. This method has been applied to practical engineering applications and has achieved good results.

In the prior art "Map Matching Algorithm in GPS Navigation System", weights are used to select matching road sections, and global positioning system (GPS) track points are projected onto the matching road sections, and vehicle track points are eliminated at intersections with parallelogram matching criteria The error along the road direction is updated through dynamic deviation to solve the deviation problem caused by factors such as satellite change, atmospheric cloud cover, multipath effect and other factors. The GPS vehicle system that introduces this algorithm is tested on a sports car in Hefei, and the results show that it can still correctly match complex road conditions and truly reproduce the driving conditions of the vehicle.

It can be seen from the above that most of the existing technologies use map matching to correct errors of the navigation system and improve the accuracy. Map matching mainly solves: finding the road the vehicle is currently driving and projecting the current positioning point onto the road the vehicle is driving on. Although the existing map matching algorithm has a high matching accuracy rate, there are still the following deficiencies:

(1) The selection of the driving road will affect the performance of the algorithm, and there are obvious differences in the performance of the algorithm for different road sections;

(2) It takes a lot of time. Due to reasons such as GPS errors, there is a large time loss in the process of finding a matching road;

(3) GPS energy consumption is high, and the GPS update speed ranges from 1 to 50hz, so the real-time performance advantage is not obvious;

(4) Map matching is obviously difficult in the case of curves. If the speed of the vehicle is from 50 to 100km/s and the GPS update speed is 1hz, the gap between the front and rear GPS points may range from 13.9 to 27.8m. This matching difficulty is even more significant for large road sections.

The current methods include video correction and topology-based pattern recognition algorithms. However, due to the need for additional acquisition equipment and calculations, this kind of correction has low portability and high cost budget. Moreover, generally speaking, on straight roads and non-complex road sections, the current GPS positioning accuracy can basically meet the navigation needs, and it is often the curved road sections and complex road conditions that need to urgently improve the accuracy requirements.

Therefore, in view of the above technical problems, it is necessary to provide a smart phone-based map matching method.

Contents of the invention

In view of this, the object of the present invention is to provide a smart phone-based map matching method, which uses the smart phone used by people daily as a platform tool, detects whether the vehicle is at a curve by detecting the motion state of the vehicle, and combines the existing The navigation system corrects its positioning error and improves the navigation accuracy.

In order to achieve the above object, the technical solutions provided by the embodiments of the present invention are as follows:

A method for map matching based on smart phones, said method comprising the following steps:

S1, install the APP with data collection function on the smart phone;

S2. Fix the smart phone in the vehicle, and open the APP with data collection function;

S3. Driving the vehicle on straight roads and curves, and manually marking events to obtain acceleration sensor data;

S4. Obtain the corresponding relationship between the smartphone coordinate system and the vehicle coordinate system;

S5. Correcting the acquired acceleration sensor data;

S6. Perform training and classification on the marked and corrected acceleration sensor data to obtain a road discrimination model;

S7. Collecting the measured road condition data, judging the road category according to the road discrimination model, and realizing map matching in combination with road topology information.

As a further improvement of the present invention, the data collected by the APP in the smart phone includes GPS data and acceleration sensor data.

As a further improvement of the present invention, the acceleration sensor data includes linear accelerations in three directions: the tangential direction of the vehicle running, the horizontal tangential direction, and the upward direction perpendicular to the horizontal plane.

As a further improvement of the present invention, the GPS data includes longitude, latitude, and altitude.

As a further improvement of the present invention, the GPS data and the acceleration sensor data also include the system time of the mobile phone and the time from the start of the mobile phone to data acquisition.

As a further improvement of the present invention, the step S6 is specifically:

The supervised learning method is used to train the acceleration sensor data and form a road discrimination model. After the model is established, the test data is used to test the accuracy and robustness of the road discrimination model.

As a further improvement of the present invention, the step S6 includes:

S61. Data preprocessing, performing desmear and denoising processing on the data;

S62. Feature extraction, extracting time-domain features and frequency-domain features of the data;

S63. Establish a road discrimination model by using a supervised learning method.

As a further improvement of the present invention, the time domain features include the preprocessed acceleration value, the mean value and variance of the acceleration value in each small segment, the mean value and variance of the entire segment of data; the frequency domain feature includes the preprocessed acceleration value conversion Amplitude values to the frequency domain, mean and variance of the amplitude values.

As a further improvement of the present invention, the road categories of the road discrimination model include obvious left curve, obvious right curve, straight road, unobvious left curve, and unobvious right curve.

The present invention has the following beneficial effects:

The present invention uses mobile phone sensors to detect curves and correct existing navigation systems, which can compensate the accuracy of civilian GPS systems and the inaccuracy of map systems to a certain extent, especially in curved road sections where driving behaviors frequently occur and driving behaviors are relatively high. With rich and complex road sections, this compensation can provide more accurate navigation services and make driving behavior safer.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments described in the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

FIG. 1 is a specific flow chart of the smart phone-based map matching method of the present invention.

2a and 2b are respectively schematic diagrams of the coordinates of a smart phone and a vehicle in a specific embodiment of the present invention.

Fig. 3 is a schematic diagram of raw acceleration sensor values including error data and noise data in a specific embodiment of the present invention.

Fig. 4 is a schematic diagram of acceleration sensor values after data preprocessing to remove dirt and noise in a specific embodiment of the present invention.

Fig. 5 is a schematic diagram of a template of a classification method in a specific embodiment of the present invention.

Detailed ways

In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described The embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

The invention is a method for correcting a navigation map based on the detection of curves on a smart phone platform. The smart phone sensor used is the built-in acceleration sensor and GPS receiving module inside the phone. Firstly, through the analysis of the data collected by the acceleration sensor, it is judged whether the vehicle has passed the curve and the situation of the curve and other information. Next, by combining the road topology information, the matching between the digital map and the road (curve or straight road) where the vehicle is located is realized, in order to correct the GPS positioning error and the lack of map information.

With reference to shown in Fig. 1, the present invention discloses a kind of map matching method based on smart phone, comprises the following steps:

1. Driving road identification:

S1, install the APP with data collection function on the smart phone;

S2. Fix the smart phone in the vehicle, and open the APP with data collection function;

S3. Driving the vehicle on straight roads and curves, and manually marking events to obtain acceleration sensor data;

S4. Obtain the corresponding relationship between the smartphone coordinate system and the vehicle coordinate system;

S5. Correcting the acquired acceleration sensor data;

S6. Perform training and classification on the marked and corrected acceleration sensor data to obtain a road discrimination model. Step S6 specifically includes:

S61. Data preprocessing, performing desmear and denoising processing on the data;

S62. Feature extraction, extracting time-domain features and frequency-domain features of the data;

S63. Establish a road discrimination model by using a supervised learning method.

2. Curve topology information matching and vehicle map correction:

S7. Collecting the measured road condition data, judging the road category according to the road discrimination model, and realizing map matching in combination with road topology information.

The present invention will be further described below in combination with specific embodiments.

1. Judgment of driving road type

S1. Install the APP with data collection function on the smart phone.

Write a data collection APP based on the smartphone platform, mainly including the data collection start button and the collection stop button, and display the road type: obvious curve starts, obvious curve ends, non-obvious curve starts, non-obvious curve ends, straight road starts and the end of the straight.

The collected data includes GPS data and acceleration sensor data. The GPS data format is shown in Table 2, and the acceleration sensor data is shown in Table 1. Among them, acceleration is the key to judge the state of the vehicle next. It uses the principle of inertia to sense the change of acceleration force. Acceleration force is the force acting on the object during the acceleration process, such as shaking, falling, rising, falling and other moving changes can be converted into electrical signals. The GPS signal includes positioning, speed measurement and high-precision time standards. The data we collect is the longitude, latitude, altitude in GPS, mobile phone system time, and the time from the beginning of the mobile phone to data acquisition.

Table 1 Acceleration sensor data and time

Table 2 GPS data and time

S2. The smart phone is fixed in the vehicle, and the data collection program is started.

a. The mobile phone is placed next to the driving seat in the car and fixed with a fixing device. For the convenience of handling, the mobile phone is fixed on the car with double-sided tape to ensure that the mobile phone and the vehicle do not move relative to each other. Do the same during the movement Three-dimensional space movement;

b. Use the built-in three-axis gravity accelerometer of the smart phone to measure the tangential direction of the vehicle, the horizontal tangential direction, and the linear acceleration perpendicular to the horizontal plane.

In this embodiment, the value obtained by the triaxial gravitational accelerometer can be obtained by the difference of the triaxial velocity meter, and can also be obtained by the second difference of the triaxial displacement meter.

Among them, the three-dimensional coordinate system of the vehicle is shown in Figure 2b. The direction connecting the driver's seat and the passenger's seat and pointing to the passenger's seat is set as the positive direction of the X-axis, and the direction along the car body and pointing to the front of the car is the positive direction of the Y-axis, which is perpendicular to the chassis and passes through the intersection of the X-axis and the Y-axis. Pointing to the sky is the positive direction of the Z axis. The three-dimensional coordinate system of the mobile phone is shown in Figure 2a along the direction of the long side of the mobile phone and pointing to the direction of the earpiece. The direction of the Y axis is the direction where the mobile phone is placed on the table. The direction is the positive direction of the X axis, and the direction perpendicular to the mobile phone screen and pointing to the sky is the positive direction of the Z axis.

S3. Driving the vehicle on straight roads and curves, and manually marking events to obtain acceleration sensor data.

When driving under different road conditions, use the APP in the smartphone to collect road data, and mark the road conditions such as straight roads, obvious curves, and unobvious curves. Specifically, the marking types in this embodiment include: obvious left curve, obvious right curve, straight road, unobvious left curve, and unobvious right curve.

S4. Obtain the corresponding relationship between the coordinate system of the smartphone and the coordinate system of the vehicle through a simple test.

This embodiment refers to an article "Sensing vihicle dynamics fordetermining driver phoneuse" published by MOBISYS in 2013. The method used by the author in this article is to apply both the acceleration sensor and the gyroscope sensor. Specific steps are as follows:

a. When the vehicle is stationary, the mobile phone can detect and record the acceleration at this time on the vehicle. The triaxial acceleration vector value obtained at this time is equivalent to the gravitational acceleration of the vehicle on the ground. Normalize the obtained acceleration vector to obtain a column vector

b. Start the vehicle without adjusting the steering wheel to ensure that the vehicle moves forward in a straight line. Record the process of the car accelerating forward, and normalize the obtained acceleration vector. This step is equivalent to obtaining the unit vector along the positive direction of the Y axis in the car coordinates, which is recorded as a column vector

c. According to the right-hand rule, the unit column vector in the positive direction of the X-axis can be obtained The transformation matrix composed of these three vectors is:

$\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; =</span>\left[\begin{array}{ccc}\stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; C</span>}}& \stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; B</span>}}& \stackrel{<span\; class="notranslate">\; ^</span>}{\mathrm{<span\; class="notranslate">\; A</span>}}\end{array}\right]<span\; class="notranslate">\; =</span>\left(\begin{array}{ccc}\mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; 3</span>}& \mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; 2</span>}& \mathrm{<span\; class="notranslate">\; x</span>}\mathrm{<span\; class="notranslate">\; 1</span>}\\ \mathrm{<span\; class="notranslate">\; the\; y</span>}\mathrm{<span\; class="notranslate">\; 3</span>}& \mathrm{<span\; class="notranslate">\; the\; y</span>}\mathrm{<span\; class="notranslate">\; 2</span>}& \mathrm{<span\; class="notranslate">\; the\; y</span>}\mathrm{<span\; class="notranslate">\; 1</span>}\\ \mathrm{<span\; class="notranslate">\; z</span>}\mathrm{<span\; class="notranslate">\; 3</span>}& \mathrm{<span\; class="notranslate">\; z</span>}\mathrm{<span\; class="notranslate">\; 2</span>}& \mathrm{<span\; class="notranslate">\; z</span>}\mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right)<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Multiply the data obtained from the accelerometer to the right Get the required motion state in the real three-dimensional space of the vehicle.

S5. Correct the acquired acceleration sensor data, and apply the correction method in S4 to correct the sensing data obtained in S3.

S6. Perform training and classification on the marked and corrected acceleration sensor data to obtain a road discrimination model.

Using supervised learning methods, training data and forming models. After the model is established, test the data to verify the accuracy and robustness of the model.

Specifically divided into the following steps:

S61. Data preprocessing:

In order to make the model more separable, a preprocessing process is required before training. Specific information processing methods: information processing starts from the original data, coordinate correction, noise removal, error data removal, effective feature information extraction, and road category classification , digital map road information extraction, road matching, etc.

Data preprocessing mainly realizes two functions: coordinate correction of data and filtering out noise and dirty data. The data obtained from the mobile phone sensor is data with noise and some random errors. Even if the mobile phone is still and flat on a horizontal surface, noise and random error data will also be generated.

In order to improve the accuracy of the system as much as possible and reduce the misjudgment caused by wrong data, these data need to reduce noise and filter out wrong data. The measures taken by the present invention are to firstly remove the wrong data, and then use smoothing filter to reduce the noise of the data. Since the percentage of error data in the overall data is not high, the obtained data can be sorted first to remove the M values with the largest absolute value (in this embodiment, M is set to 10 in the specific experiment, because each intercepted The data length is 150 data, and the sampling frequency of the acceleration sensor is set to 50hz, then removing 10 data out of every 150 data will not have a big impact on the judgment of the data itself, but it can filter out the wrong data to avoid a misjudgment). After that, the obtained data can be average-filtered to achieve the desired effect of denoising and de-dirty data.

The effect before and after treatment can be compared, as shown in Figure 3 and Figure 4. The data before processing contains loud noise and some obvious wrong data. The noise of the processed data is significantly reduced, the error data is removed cleanly, and the information of the data itself is not greatly affected.

S62, feature extraction:

The extracted features include time domain features and frequency domain features.

Time-domain features include: preprocessed acceleration values, the mean and variance of acceleration values in each segment (10 to 20 values can be set to a segment), and the mean and variance of the entire segment of data. The frequency domain features include: the amplitude value converted from the preprocessed acceleration value to the frequency domain, and the mean value and variance of the amplitude value.

Then, determine the road conditions by selecting a feature mark template: obvious curves, straight roads, and unobvious curves. In the actual template selection, the curves are further divided into left turns and right turns, and the cases where the curves are not obvious are also divided into left turns and right turns. In the present invention, the curve categories are actually classified into five categories: obvious left curve, obvious right curve, straight road, unobvious left curve, and unobvious right curve.

S63. A classification model is established by a supervised learning method.

The classification method adopted in this embodiment does not undergo dynamic time rounding, and uses the template in FIG. 5 . The data in Figure 4 are all collected by experienced drivers during multiple driving on straight roads and curves, and erroneous data are eliminated.

2. Curve topology information matching and vehicle map correction

After the template is selected, the road conditions can be determined through the collected acceleration sensor data: the curve is obvious, the straight road is not obvious, and the curve is not obvious, and then the digital map road topology needs to be analyzed.

Road information is composed of a series of GPS points. GPS information includes longitude data and latitude data. According to GPS information, longitude and latitude are regarded as two-bit information, and the correlation function of longitude and latitude is formed through curve fitting. Then calculate the slope on the GPS point through the fitted curve, that is, the tangent slope of the curve on the GPS point, after passing the arc tangent, it is the tangent angle.

The specific matlab code is as follows:

In this way, the geometric information on the road can be obtained, and it is easy to know whether the road has obvious left curves, obvious right curves, straight roads, unobvious left curves, or unobvious right curves by judging the angle. Combined with information such as map road connectivity, the information of the digital map can be implicitly mapped to the dimensions containing road connectivity information and road curve information.

Then, the curve information analyzed from the acceleration sensor obtained on the smartphone is matched with the curve information on the map. If the information after the map conversion is the same as the information judged by the sensor, then the current GPS point is matched to the current road section, otherwise, the detection and search are performed again.

To sum up, the present invention uses the mobile phone sensor to detect curves and correct the existing navigation system, which can compensate the accuracy of the civilian GPS system and the inaccuracy of the map system to a certain extent, especially on curved roads where driving behavior frequently occurs. Such compensation can provide more accurate navigation services and make driving behavior safer.

It will be apparent to those skilled in the art that the invention is not limited to the details of the above-described exemplary embodiments, but that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Accordingly, the embodiments should be regarded in all points of view as exemplary and not restrictive, the scope of the invention being defined by the appended claims rather than the foregoing description, and it is therefore intended that the scope of the invention be defined by the appended claims rather than by the foregoing description. All changes within the meaning and range of equivalents of the elements are embraced in the present invention. Any reference sign in a claim should not be construed as limiting the claim concerned.

In addition, it should be understood that although this specification is described according to implementation modes, not each implementation mode only contains an independent technical solution, and this description in the specification is only for clarity, and those skilled in the art should take the specification as a whole , the technical solutions in the various embodiments can also be properly combined to form other implementations that can be understood by those skilled in the art.

Claims (9)

1. based on a map-matching method for smart mobile phone, it is characterized in that, said method comprising the steps of:

S1, smart mobile phone is installed there is the APP of data acquisition function;

S2, smart mobile phone is fixed in vehicle, opens the APP with data acquisition function;

S3, steering vehicle travel forthright and bend, and manually carry out event mark, obtain acceleration transducer data;

The corresponding relation of S4, acquisition smart mobile phone coordinate system and vehicle axis system;

S5, to obtain acceleration transducer data correct;

S6, the acceleration transducer data marking and correct are carried out to training classification, obtain road discrimination model;

S7, collection actual measurement road condition data, judge category of roads according to road discrimination model, and in conjunction with road topology information realization map match.

2. method according to claim 1, is characterized in that, the data that the APP in described smart mobile phone gathers comprise gps data and acceleration transducer data.

3. method according to claim 2, is characterized in that, described acceleration transducer data comprise vehicle and travel tangential direction, horizontal tangent direction and perpendicular to the linear acceleration on surface level upward direction three directions.

4. method according to claim 2, is characterized in that, described gps data comprises longitude, latitude, sea level elevation.

5. method according to claim 2, is characterized in that, described gps data and acceleration transducer data also comprise cell phone system time and mobile phone to time of data acquisition.

6. method according to claim 1, is characterized in that, described step S6 is specially:

Adopt supervised learning method, training acceleration transducer data also form road discrimination model, after model has been set up, and the accuracy of test data inspection road discrimination model and robustness.

7. method according to claim 1, is characterized in that, described step S6 comprises:

S61, data prediction, go dirty and denoising to data;

S62, feature extraction, extract temporal signatures and the frequency domain character of data;

S63, employing supervised learning method establishment road discrimination model.

8. method according to claim 7, is characterized in that, described temporal signatures comprises the average of accekeration in pretreated accekeration, each segment and variance, the average of whole segment data and variance; Frequency domain character comprises pretreated accekeration and is transformed into the range value of frequency domain, the average of range value and variance.

9. method according to claim 7, is characterized in that, the category of roads of described road discrimination model comprise obvious, the right bend of left bend obviously, not obvious, the right bend of forthright, left bend is not obvious.