RU2663692C2 - Method and system of determining current and recommended lane of motor vehicle - Google Patents

Abstract

FIELD: navigation system.SUBSTANCE: invention relates to the field of navigational instrument making and can find application in navigation systems of vehicles. Method for determining the current lane is based on the application of a moving vehicle obtained from the global navigation satellite system of coordinates and the so-called "digital marking". "Digital marking" is a mathematical model of individual sections of the road-street network, built on the basis of a set of observations on the movement of many other vehicles that previously passed on this section of the road, and describes the actual location of the lanes on the carriageway, with respect to the selected reference hyperplanes. Definition of the lane that the vehicle occupies is of a probabilistic nature. To improve the accuracy of the calculation, a sequence of observations corresponding to different, but close to each other, time points can be used. In this case, the method makes it possible to estimate the relative utilization of individual bands on sections of the road-street network, as well as to detect overlapped bands due to repair work or road accidents. Introduced concept of "micronavigation" is used, which is a process of developing and visualizing recommendations for timely rebuilding in the right lane while following a predetermined route in order to successfully perform maneuvers and turns at intersections and traffic junctions, obstacle clearance, etc. In addition to the method for determining the lane, variants of implementing the software and hardware system are proposed, providing the ability to calculate the current lane and make recommendations for changing it both with access via global communication networks (online mode) and without such access (offline mode).EFFECT: expanded functionality by means of improving the quality of vehicle navigation and improving information services for users using navigation software and devices.13 cl, 6 dwg

Description

FIELD OF THE INVENTION

The invention relates to a navigation system of a vehicle and, in particular, to a method for probabilistically determining the current lane and forming recommendations but timely restructuring when following a given route.

State of the art

Currently, there are a large number of navigation programs designed for use in vehicles and serving as an assistant in choosing a route and following it. Similar navigation programs are installed on mobile user devices (tablets and smartphones), stand-alone devices, as well as in computers built directly into the car. The most popular of these programs are Yandex. Navigator (from the Yandex company), Google Maps (from the Google company), Navitel (from the Navitel company), CityGUID (from the MIT company) and others. As a rule, these programs have similar functionality - building routes from one point to another and maintaining along a given route, displaying the current speed, calculating the time of arrival at the end point. At the same time, the route is a set of interconnected major key points of the road graph (intersections, junctions, multi-level junctions, etc.).

The calculation of the current position on the route is based on obtaining vehicle coordinates from global satellite positioning systems: GPS (USA), GLONASS (RF), Gallileo (EU). Navigation devices installed in vehicles have hardware support for one or several similar systems. The most common to use are devices that support GPS or GPS and GLONASS navigation systems at the same time.

It should be noted that at present, all navigation programs “tie” the obtained coordinates of a moving vehicle to a single segment (edge) connecting the nodes of the road graph. Visualization of the route and the current location of the car on the screen of the navigation program is presented, as a rule, in the form of a wide band and a large marker, respectively. At the same time, for better visual perception, the width of the strip corresponds to the width of the entire or almost the entire roadway. In other words, there is no information about traffic on a particular lane. This problem is especially relevant in large cities, where there are a large number of multi-lane roads and complex transport interchanges. Many navigation programs are sometimes misleading or provide incomplete and untimely information, which often leads to the inability to carry out the necessary maneuver (turn, turn, etc.) and leave the route. The consequence of leaving a given route is the need to rebuild it, which entails a significant loss of time and an increase in vehicle mileage.

The Google Maps program has the ability to display information about lanes in front of some intersections. The program in graphical form (in the form of a sign) displays the lane that must be taken to complete the maneuver. However, at the same time, it does not provide and does not take into account information about which lane the vehicle is currently occupying.

In addition to the above, an important factor for a successful and safe movement is the timely receipt of information on repairs carried out along the route, traffic accidents and other situations that somehow affect the driver’s different decisions. Some existing programs allow you to manually set a warning message for other drivers, indicating the approximate location. However, the gradation of bands used in this case (of which there are usually three — the “left row”, “middle row”, “right row”) is not always sufficient. On roads with more than three lanes, the term "middle row" may indicate several adjacent lanes at once. This fact does not allow the driver to make a timely decision on the need to change lanes in the adjacent lane.

An alternative way to determine the position of the vehicle relative to the road marking lines is described in the patent of the Russian Federation for invention RU 2422772 C1, which is based on the use of other physical principles - measuring the distance from the road marking line to the vehicle and the angle of deviation of the longitudinal axis of the vehicle relative to the road marking lines using optical location blocks. This method can work in a limited number of cases, for example, when the markup is present and it is visible visually. In winter, when the roads in most regions are covered with snow, this method will not work.

Another patent close to this application is RU 2404409 C1. However, it should be noted that it only describes a map display device for a vehicle and provides a method for visually displaying lanes on a map in a navigation device, without directly disclosing the principle of formation of "digital marking".

SUMMARY OF THE INVENTION

This invention is intended to solve the above problems associated with the determination and visualization of the number of lanes on the roadway, the current lane on which the vehicle is located, as well as recommendations aimed at timely lane change to other lanes. The proposed solution is based on the processes of collecting, storing and processing data corresponding to a large number of observations from many vehicles with a navigation device installed in them.

Consider these processes in more detail. Both determining the number of lanes, the current occupied lane, and recommendations for changing it are calculated on the basis of a set of data (observations) previously obtained from other vehicles and processed using a combination of mathematical algorithms. At the same time, each individual section (segment) of the road-street network is characterized by its own set of values of the mathematical model, which can be called "digital marking". This mathematical model describes the actual location of the lanes on the carriageway of a given section with the accuracy necessary to adequately calculate the lane occupied by the car.

The key idea of the proposed method is to convert the coordinates of the vehicle’s location, which are a pair of latitude-longitude values, into one value — the distance to the reference hyperplane corresponding to this segment of the road. In this case, a reference hyperplane is either a geographical point with known coordinates, or a conditional straight or curved line passing through a given point and having the following property - a vehicle moving within a given road segment along one lane (without rebuilding) maintains its distance to this hyperplane (see Fig. 1). An example of a hyperplane is the dividing line on a straight section of the road or the center of a curve (turn).

Note that the term "reference hyperplane" is in no way associated with a two-dimensional geometric object in three-dimensional space. This name is borrowed from the theory of machine learning, where the concept of a hyperplane is used to designate a separating space of attributes whose dimension is less than the dimension of the original space. In this case, the reference hyperplane is used to reduce the space of signs — two signs that specify the coordinates of the car (latitude and longitude) are replaced by one — the distance to the reference hyperplane.

The data on the coordinates of the vehicle, its speed and direction, obtained from the global navigation system, are inaccurate due to many factors. Such factors include: weather conditions, the presence or absence of residential buildings, the terrain, the current position of the satellites, rounding errors during the calculation, etc. In other words, the received data contains a useful signal and some random noise component.

Consider the special case when the carriageway contains only one lane for traffic. It can be assumed that, by analogy with many other physical processes occurring in nature, the observed data (i.e., the distance from passing vehicles to the reference hyperplane) will be described by a normal distribution (Gaussian distribution), and its mathematical expectation will correspond to the center of the strip in question. Extending this statement to the case of a multi-lane road, we can say that the observed distribution of distances to the reference hyperplane in this case will be the sum (mixture) of normal distributions, each of which describes movement in a separate lane (see Fig. 2). Moreover, in the general case, the weight of each of the distributions in the total amount will be different. For example, lanes allocated for public transport will have little weight relative to neighboring lanes. Shown in FIG. 2 distribution is very schematic and is intended only to explain the basic principles underlying the proposed method. In reality, this distribution will look much more complicated. The values of distances, where normal distributions "replace" each other, are the boundaries of decision-making. When using a sufficient number of independent observations in the calculations, these decision boundaries will correspond to real road markings that separate lanes. In fact, these border values will constitute the “digital marking” of the road segment under consideration. Of course, this mathematical model is greatly simplified, because it does not take, for example, the movement of the vehicle relative to the center of the lanes, frequent rearrangements from one lane to another, the location of the navigation receiver inside the passenger compartment, and other factors. In connection with the above, the solution to the problem in question is probabilistic. The proposed method does not guarantee absolute accuracy in determining the current lane. As an answer, it only returns probabilities that correspond to movement in a particular band.

Dividing road segments into separate “digital lanes” opens up opportunities for improving the quality of information services related, for example, to navigation along a given route and reporting traffic jams. In particular, the average speed can not be calculated for the entire segment or its part, but for individual lanes. In this regard, it becomes possible to detect short-term overlap in separate lanes (due to a malfunction of the vehicle or a traffic accident). Such a detection is possible if there is sufficient traffic and received data from passing cars, as well as with frequent rebuilding of the mathematical model of a road segment and recalculation of the average speed in separate lanes. Typically, an overlapped strip will be characterized by:

- a sharp decrease in the average speed in this lane in time;

- lower average speed compared to other lanes in this segment.

To determine the long-term overlap of lanes (for example, due to ongoing repairs), a comparison of several “digital markings” of the same road section for different periods of time can also be used. The overlapped band will be distinguished by a sharp decrease in the density of observations corresponding to this area.

In addition, the "digital stripes" open up prospects for micronavigation when driving along a given route. Unlike global navigation, where navigation is carried out through separate nodes of the road-street network (intersections, multi-level traffic interchanges, etc.), micronavigation is used to determine the need for timely lane changing. On multi-lane roads, different lanes are used to travel on different routes. Timely rebuilding in a particular lane is an important condition for the correct movement on the selected route. Modern navigation programs do not warn of the need to change lanes, which often leads to the inability to carry out the necessary maneuver (turning, turning, etc.) and leaving the route. The consequence of leaving a given route is the need to rebuild it and move along a new route, which entails a significant loss of time for the car owner and an increase in vehicle mileage.

The ability to track the current lane allows you to timely warn the driver about the need for a lane. In this case, the system is similar to recommendation systems - it takes its decision on the basis of “experience” obtained from data from other vehicles that previously performed the given maneuver. If an anomaly is detected when comparing the current position of the car and the mathematical model that describes the desired maneuver on this segment of the road, the computing module gives recommendations to change the lane. In this case, it should be noted that the accuracy of determining the current band in this case is not important. For a recommendation, it is enough to identify an anomaly, that is, a significant deviation of the current distance to the reference hyperplane from the totality of the previously obtained values.

Visualization of the location of the vehicle relative to the lanes, congestion detected, as well as recommendations for changing the lane, may include both the output of information in graphical form on the screen of the navigation program, and in the form of voice messages.

Brief Description of the Drawings

The invention is further explained in the description of specific options for its implementation with reference to the accompanying drawings, in which:

- FIG. 1 describes the general principles underlying the determination of a lane occupied by a vehicle.

- FIG. 2 describes the concept of a supporting hyperplane.

- FIG. 3 describes the composition and interaction scheme of system components when working in online mode.

- FIG. 4 describes the composition and interaction scheme of system components when working offline at the stage of planning the route of movement.

- FIG. 5 describes the composition and interaction scheme of system components when working offline while the vehicle is in motion.

- FIG. 6 describes the composition and interaction scheme of system components when working offline after a trip.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

This section provides a detailed description of two embodiments of the invention. The first option assumes the presence of a network connection between the navigation device and the server part of the system that functions during the vehicle’s movement. In this case, all calculations are performed on the server side, the navigation device only visualizes the obtained calculation results. An alternative option involves the placement of a computing module as part of a navigation device, which allows you to use the proposed method for determining the current band in places where connection to the server is unavailable.

While driving on the road, the navigation equipment installed in the vehicle receives data from the satellites of the global navigation system and calculates the coordinates of the current position of the vehicle, its instantaneous speed and direction of movement. This set of data relating to one moment in time and characterizing the movement of the vehicle at this point in time will be called observation below. The calculated data using the global communication network (in particular, the Internet) is transmitted to the data center on one of the servers of the information processing subsystem (POI). To increase the throughput of the system, ensure its fault-tolerant operation and reduce the delay time for the transmission and processing of information, several data centers can be used that are geographically remote from each other.

In addition to the information processing subsystem, the data center also includes: a storage subsystem (HR), a computational module (VM), a modeling subsystem (PM), and a model database (PM). It should be noted that this composition of the subsystems is very conditional. It is possible to use various combinations of these subsystems within the same data center. For example, one data center may be entirely responsible for storing raw data and modeling, and another for storing ready-made mathematical models and calculating the probability of lanes of vehicles currently moving.

The following is a more detailed description of the basic principles of interaction between these subsystems, as if they were located in one data center.

The processing subsystem simultaneously transmits the data received from the vehicle to the storage subsystem (stage 2) and the computing module (stage 3). The data packet may include one or several observations at the same time, relating to one or more points in time, respectively. Storage of data on the coordinates, instantaneous speed and direction of movement of vehicles is necessary for the initial calculation of the mathematical model of individual segments of the road-street network (“digital marking”), as well as its further refinement. The computing module accesses the database of models (stage 4), which returns the mathematical models of the segments of roads along which the vehicle is moving (stage 5). The computing module calculates the probability of movement in a particular lane and, if necessary, makes recommendations for rebuilding. The generated response is returned to the information processing subsystem (stage 6), from where they are transmitted back to the user device via the global communication system as part of the data packet (stage 7), where visualization takes place - by means of graphic display and / or voice notification.

The modeling subsystem is responsible for the formation of relevant “digital markup”, which regularly requests “raw data” in the storage subsystem (stage 8), receives this data (stage 9), and builds mathematical models of individual road segments. The frequency of updating the model can be different and depends on such factors as: the capacity of the road segment, the frequency of road traffic accidents, ongoing repairs related to overlapping lanes, etc.

The quality of the mathematical model for each segment is determined by the number of measurements obtained from vehicles moving along these segments. In the absence of measurements, it is impossible to build a mathematical model of the segment, and, therefore, to determine the current strip. With the accumulation of a certain threshold number of observations, the “digital marking” will correspond with real accuracy with great accuracy. This accuracy will be achieved, despite the random errors inherent in each such observation, which is achieved due to the effect of big data. The accuracy of determining the coordinates depends on many factors - weather conditions, terrain, the presence of buildings, etc. Therefore, the threshold value for each segment will be different.

The generated models are stored in the database (stage 10), from where they are retrieved by the computing module upon request.

The number of observations in the data packet and the transmission interval may be different depending on the accuracy shown and the power consumption of the device. The least accurate and at the same time energy-consuming will be the option of transmitting one coordinate pair (latitude-longitude) once every few seconds. In the case of transferring in a single packet several pairs of coordinates received every second, with an interval of 5 seconds, the accuracy of determination should increase significantly. It is possible to transmit coordinate pairs with overlapping, for example, every 5 seconds a data packet containing observations of the previous 10 seconds of movement is transmitted. It is assumed that the operating mode will be selected by the manufacturer of the navigation program. In this case, the interval should be chosen so that the probability of changing from one lane to another over a given period of time is minimal.

An alternative is to use a computing module as part of the software on a user device. In this case, the "digital marking" is loaded in advance when drawing up the route of movement or on request when approaching a segment of the road. However, the need to transfer vehicle coordinates remains due to the need to refine the mathematical model of the road segment. In this case, the requirement for efficient data transfer can be mitigated. The transfer and processing of all or part of the track can be carried out in batch mode later, when it becomes possible to transmit data about the movement of the car through the available communication networks.

The interaction of the navigation device with other components of the system can be divided into three stages - the stage of loading the route (see Fig. 4), movement along the route (see Fig. 5) and unloading the track (see Fig. 6). At the stage of loading the route, the navigation device through the global communication network accesses the information processing subsystem located in the data center (stage 1 in Fig. 4). The data processing subsystem turns to the database of mathematical models (stage 2), which returns the “digital markings” of road segments corresponding to the proposed route of movement (stage 3). These mathematical models are downloaded via the global communication network to the navigation program on the user device (stage 4). While the car is following the route, the computing module, which is part of the navigation device, makes the necessary calculations related to the calculation of the current lane and the development of recommendations for restructuring (stage 1 in Fig. 5). Given that the "digital markup" has already been downloaded earlier, access to the data center through communication channels is not required. After the trip is completed or in other cases when access to the communication network appears, the navigation device transmits the observation accumulated during the movement of the vehicle (stage 1 in Fig. 6). Observation data is stored in the information storage subsystem (stage 2). The modeling subsystem regularly requests “raw” data in the storage subsystem (stage 3), and uses the obtained data to refine mathematical models (stage 4). Refined “digital markings” are stored in the database of models (stage 5), from where they are later reloaded into users ’navigation devices.

Claims (12)

1. The method of determining the lane on the carriageway occupied by the vehicle, which consists in converting the coordinates of the navigation device installed in the vehicle received from the satellite navigation system to the distance to the reference hyperplane, which is a conditional line passing through a geographical point with known coordinates and having the property of maintaining the distance from the vehicle to this hyperplane for the transpo moving within the given segment of the road merchant means in one lane without performing a rebuilding maneuver, comparing this distance with decisive boundaries corresponding to the mathematical model (“digital marking”) of a road-street network segment obtained by choosing the optimal distribution function that most closely matches the set of observations about the movement of other vehicles, previously on this segment of the road, calculating the probability of movement on a particular lane and choosing the most likely option. 2. The method of determining the lane on the roadway occupied by the vehicle, which consists in converting the coordinates of the navigation device installed in the vehicle received from the satellite navigation system into a distance to a geographical point with known coordinates, which has the property of maintaining the distance from the vehicle to this point for a vehicle moving within a given segment of the road in one lane without performing a lane maneuver, comparing this states with decisive boundaries corresponding to the mathematical model (“digital marking”) of a segment of a road-street network, obtained by choosing the optimal distribution function that most closely matches the set of observations about the movement of other vehicles that were previously on this segment of the road, calculating the probability of movement for that or another band and choosing the most likely option. 3. The method according to p. 1, characterized in that for making the decision an additional set of independent observations is used, corresponding to the positions of the vehicle at various points in time in the past. 4. The method according to p. 3, characterized in that for making a decision the composition of the set of independent observations is supplemented by part of the set of independent observations used in the previous calculation, and the set of observations is used in the following calculation. 5. The method according to p. 3, characterized in that for making a decision the composition of the set of independent observations is supplemented by part of the set of independent observations used in the previous calculation, and part of the set of observations is used in the following calculation. 6. A system for determining the current lane of a vehicle, including a navigation device located in the vehicle, as well as a processing, storage, modeling subsystem, a model database and a computational module included in the data center accessible to the navigation device using global communication network, which collects and transfers to the data processing center observations on the coordinates, speed and direction of movement of the vehicle, however collecting observations received from various vehicles, calculating and storing mathematical models corresponding to various segments of the road network (“digital marking”), comparing current observations with the parameters of the mathematical model corresponding to the road segment, calculating the probabilities of movement along individual lanes of the carriageway and development of recommendations for the restructuring of the vehicle in other lanes. 7. The system according to p. 6, characterized in that the computing module, which calculates the probability of movement in individual lanes and the development of recommendations for the restructuring of the vehicle into other lanes, is part of the navigation program located in the vehicle, which makes it possible to determine the current occupied lane movement and making recommendations on the need to occupy other bands without access to a data center through a global communication network. 8. A method for determining the relative congestion of individual lanes of segments of a road-street network, which consists in dividing the totality of observations received from the navigation devices of vehicles for a certain time interval into groups corresponding to the most probable lanes for each vehicle, calculating and comparing average vehicle speeds assigned to each such group. 9. A method for determining the overlap of a separate lane on the roadway, which consists in the formation, on the basis of the totality of observations received from the navigation devices of the vehicles for a certain time interval, of the optimal distribution function of the distances from the vehicles to the reference hyperplane, most closely corresponding to the set of observations about the movement of the whole aggregates of vehicles on a section of the road-street network and representing the sum (mixture) of such normal distributions d, that the mathematical expectation of each of which corresponds to the center of a separate lane, calculating the relative values (weights) of the obtained distribution functions, comparing them with each other and choosing the one that corresponds to the distribution of distances to the reference hyperplane with a minimum relative to other distributions weight. 10. The method according to p. 9, characterized in that in addition to comparing the relative values (weights) for each distribution corresponding to a separate lane, a comparison of the dynamics of changes in relative values (weights) over time is applied. 11. A way to make recommendations to users of navigation programs on the timely change of the lane of a vehicle to successfully complete the necessary maneuver when following a given route, which consists in comparing the current lane with the lanes occupied by other vehicles that were previously on this segment of the road and performed a similar maneuver, detecting deviations above a given threshold in the results of this comparison, the formation and transmission of a data packet, and recommendation-containing machine-readable form, and visualization by displaying recommendations on the screen of the navigation software and the voice announcement. 12. A method of making recommendations to users of navigation programs on the timely change of the vehicle lane for successfully performing the necessary maneuver when following a given route, which consists in comparing the distances to the reference hyperplane corresponding to the segment of the road on which the vehicle is located with the distance to the same reference hyperplane for other vehicles previously on this segment of the road and performing a similar maneuver, detection tkloneny vyshezadannogo threshold in the results of this comparison, generating and transmitting a data packet containing a recommendation machine-readable form, and visualization by displaying the recommendation on the screen of the navigation program, and voice messaging.

Images