GB2513738A - Method for determining a position of a vehicle by a navigation system of the vehicle - Google Patents

Abstract

The invention relates to a method for determining a position x1, x2, y of a vehicle by a navigation system of the vehicle, with the navigation system receiving position data by aid of a positioning system, such as GPS, and determining the vehicles position on a map M stored in the navigation system. The navigation system receives data on a width w1 of a current road R the vehicle is travelling on from a camera system in the vehicle. It then verifies if the vehicles position on the map M is consistent with the width w1 of the current road R, and corrects the vehicles position on the map M to a new position on the map M that is consistent with the width w1 of the current road R if the vehicles initially determined position on the map it is not consistent with the width w1 of the current road R. Specific locations on the map M can be marked as checkpoints P, and the method is only executed if the vehicle is close to one of the checkpoints.

Description

Method for determining a position of a vehicle by a navigation system of the vehicle The invention relates to a message for determining a position of the vehicle by a navigation system of the vehicle according to the preamble of independent claim 1.

Many vehicles are equipped with a navigation system. Position data available in such a navigation system, for example data of the global positioning system (GPS) has only a certain, limited precision. As a consequence, the navigation system suffers from an ambiguity of matching a moving vehicle onto a correct position on a map if this matching process, usually called map-matching, is solely based on the position data. This is especially true in special circumstances, e.g. when parallel roads are very close to each other. This ambiguity can result in faulty and confusing guidance by the navigation system, frustrating users when the navigation system fails to identify the correct position on a map.

The DE 10 2010 033 729 Al describes a method for improving the precision of a navigation system of a car. There, a camera in combination with a signal of a global positioning system helps to determine in which lane of a single road a vehicle is travelling.

This is achieved by evaluating the lane marks on the road and combining them with the position data of the global positioning system.

It is an objective of the present invention to determine a position of a vehicle on a map by a navigation system of the vehicle more precisely.

According to the invention, there is provided a method for determining a position of a vehicle by a navigation system of the vehicle according to the independent claim 1.

Further advantages and embodiments are set out according to the dependent claims, the

detailed description and the figures.

The inventive method for determining a position of a vehicle by a navigation system of the vehicle comprises the steps of receiving of position data by aid of a positioning system and of determining the vehicle's position on a map stored in the navigation system. These two steps are executed by the navigation system. In order to determine the position on the map more precisely, the navigation system executes three further steps. First, the navigation system receives data on a width of a current road the vehicle is currently travelling on from a camera system of the vehicle. Here, the navigation system may itself be part of the camera system and calculate the width of a current road based on visual data. Second, the navigation system verifies if the vehicle's position on the map it has determined before is consistent with the width of the current road. Third, the navigation system corrects the vehicle's position on the map to a new position on the map that is consistent with the width of the current road if the vehicle's position on the map it has determined before is not consistent with the width of the current road. In particular, the new position on the map is on a road parallel to the road the prior position of the vehicle has been located on. So, the inventive method checks for the road width or its change with a visual sensor such as a camera for determining the correct link from a position data to a location on a map stored in the navigation system. The process of creating this link is also called map matching. Map matching on parallel roads is an ubiquitous challenge for navigation systems throughout the world. The inventive method gives the advantage that it solves the map matching uncertainty on parallel roads using only a camera system and a positioning system. The invention makes map matching more reliable in the close range and thus allows for a better orientation in the close range. So, the invention benefits a navigation system of the vehicle by improved route guidance and customer satisfaction.

Furthermore, only a low computational load is generated by the method, as a road width detection, e.g. by a camera, is computationally cheap. Furthermore, the method does not require any additional data or data mining, as the maps stored in modern navigation systems already include information on the road width.

The method can be refined so that the map stored in the navigation system has specific locations marked as check points, and the steps of receiving data, verifying the vehicle's position on the map, and correcting the vehicle's position on the map are only executed if the vehicle's position on the map is close to one of the check points, in particular if the vehicle, according to its position on the map, is passing a check point or has moved away a predefined distance from a check point after passing it. This results in the advantage that the method is executed only when it is needed, so that the computational cost of executing the method is further reduced.

In a further embodiment, the specific locations are entry points of parallel roads and/or ramps, in particular ramps to and/or from an elevated road. As in proximity of parallel roads and/or ramps there is a high probability for a mismatch of position data and position on the map, i.e. a high probability for a faulty map matching, a verification of the vehicle's position on the map can be especially useful.

The method can further be refined by comparing the width that the parallel roads or ramps have according to the map prior to the steps of receiving data, verifying the vehicle's position, and correcting the vehicle's position, and executing those stops only if the parallel roads or ramps widths are different, in particular if the differ enough so that the data from the camera system is sufficiently precise for the navigation system to recognize the difference of the widths. This gives the advantage that an unnecessary computational load is avoided in a situation where it is predictable that the method would not give satisfactory results, i.e. where the method would not enhance the map matching.

By considering the following detailed description of exemplary embodiments in conjunction with the accompanying drawings, the teachings of the present invention can be readily understood, and at least some visional specific details will appear.

Herein: Fig. 1 shows a map of an intersection with an entry to a side road parallel to one of the crossing main roads; Fig. 2 shows a map of a side road entry to a side road that is parallel to a main road; and Fig. 3 shows a flow chart of an exemplary embodiment of the inventive method.

In the Figs., identical elements or elements with an identical function are marked with identical reference signs.

Fig. 1 shows a map M of an intersection with an entry to a side road that is parallel to a main road. At the intersection at-hand, one main road in a y-direction and another main road in the x-direction are crossing each other in a right angle. In this example, on the right-hand side of the intersection, in a positive x-direction, the main road that is parallel to the x-direction splits up into a main road and a side road that runs parallel to the main road. That is, the intersection marks the beginning of a side road running parallel to the main road in the x-direction. Here, the entry point to the side road, that is, the point where the side road starts to run parallel to the main road, is marked as check point p. This mark may not be visible to a driver of a vehicle who is e.g. looking at the map M, which, in this example, is a map M of the navigation system. On the left hand side of the intersection, again on the main road that is parallel to the x-axis, there is marked the position xl of a vehicle moving along its forward direction f towards the intersection. The forward direction is depicted by an arrow here. As there is no parallel road existent close to the position xl of the vehicle, only a standard map matching procedure as known from prior art is done in this example at this the position xl of the vehicle.

As the vehicle is continuing its travel in the positive x-direction, it will cross the intersection and pass the check point p that marks, e.g., the beginning of a side road or a ramp parallel to the main road. There, the vehicle may or may not enter the side road that runs parallel to the main road in the x-direction. For conventional map matching, this is impossible to discern. However, in order to avoid faulty map matching and confusion of a driver of the vehicle, the navigation system should identify the vehicle's position xl correctly on the current road R, i.e. the road the vehicle is currently traveling on. In order to do so, in this example, a width wl of the current road R is measured. For example, this can be done a certain, predefined distance after passing check point p, for instance 20 meters after passing the check point p. If the current road R is the main road, the width wi measured should be the width that is stored in the map M for the main road. If the current road R vehicle is travelling on is the side road, the width wl measured should be the width that is stored in the map M for the side road. So, the measuring of the width wi of the current road R allows a non-ambiguous identification of the current road R and thus to identify the vehicle's position xl correctly. Consequently, the parallel roads need to differ in their width in order to allow to resolve the ambiguity of the conventional approach to map matching.

In the present example, the main road is the current road H the vehicle is traveling on, so the measured width wi of the current road H will be significantly larger than the width w2 of the side road stored in the map M. Accordingly, as the position x2 of the vehicle is consistent with the measured width wi of the current road H, the position x2 of the vehicle on the main road will be set as the actual position of the vehicle on the map M. The alternative position y of the vehicle on the side road will be discarded, even though it might, for instance, represent the vehicle's actual position according to the position data of the positioning system. So, the position of the vehicle on the map M can be determined more precisely than by conventional map matching. In this and the following example, the side road could have an altitude different from the side road and feature e.g. a ramp as well.

Fig. 2 shows a map M of a side road entry to a side road that is parallel to a main road.

Both main road and side road are parallel to the x-direction in this example. Similar to Fig. 1, the vehicle's position xl with the forward direction f is marked on the current road H in this example. So the vehicle at the vehicle's position xl is approaching the side road entry. Again, the entry point of the side road is marked as check point p here. So, e.g. if the vehicle is actually entering the side road that hence becomes the current road H of the vehicle, it will be possible that the navigation system of the vehicle does not display the vehicle's position x2 on the current road H correctly, but rather a faulty alternative position y, in this case on the main road. In order to avoid this incorrect information, the width wi of the current road H is measured in proximity of the check point p, for instance after the vehicle has passed it. This can for instance be done by a video camera of the car that is able to recognize the curbs or road edges on the side of the current road H, for example, and measure the width between left and right curbs or road edges. Once the width wi of the current road R is measured, it can be compared to the widths stored in the map M for the parallel roads close to the position the vehicle has according to the position data of the positioning system, in this example to the widths stored for the main road and the side road. Then, it can be decided unambiguously whether the correct position is the vehicle's position x2 or an alternative position y as each of the positions x2, y would result in measuring a different width wi, w2 according to the information or in the map M. Fig. 3 shows a flow chart of an exemplary embodiment of the inventive method. Here, in a first step 1, a map matching as known from the state-of-the-art is performed. Then, for example in regular intervals, possibly also continuously, it is checked whether a check point p is in the proximity of the vehicle or not. If this is not the case, step 1 is continued. If the vehicle is approaching or passing by a check point p, however, a further checking is done in a next step 3 of the method. In the present example, it is checked whether the vehicle already has moved away from the check point p more than a pregiven distance, meters here. Here, an arbitrary distance different from 20m, including Om, may be used. If the vehicle has not reached the pregiven distance, step 3, that is, the checking of the distance between the vehicle and the check point p is repeated. When the vehicle has moved away from the check point p more than the pregiven distance, e.g. 20 meters, the data stored in the map M (Figs. 1, 2) is checked with regards to the widths of the closest parallel roads in step 4 of the method. That is, the navigation system of the vehicle checks if the two parallel roads that exist in the vicinity of the vehicle have, according to the information stored in the map of the navigation system, the same width or not. As the check point p is only set in locations where parallel roads actually exist, the checking of the widths will always give a reasonable result. If the two parallel roads do have the same width, the method is aborted and the map matching of step 1 according to the prior art is continued. If the parallel roads do not have the same width, however, another step 5 is executed. Within the step, the width wi (Figs. 1, 2) of the current road R (Figs. 1, 2) is measured by identifying e.g. road edges and/or curbs on the side of the road and measuring the distance between the left and right curbs and/or road edges. This can be done by a state-of-the-art video camera system with image recognition capability. After the width wi (Figs. 1, 2) of the current road R (Figs. 1, 2) has been determined, the data on the width wi (Figs. 1, 2) is received by the navigation system which then can the in a final step 6 verify if the position of the vehicle according to the position data is consistent with the width wi (Figs. 1, 2) of the current road R (Figs. 1, 2) by comparing the road width at different possible positions of the vehicle with the width wi (Figs. 1, 2) of the current road R (Figs. 1, 2). The step 6 also includes the final step of the map matching, that is correcting the position of the vehicle on the map to a vehicle's position xl (Figs. 1, 2) on the map that is consistent with the width wi (Figs. 1, 2) of the current road R (Figs. 1,2).

So, the method can be referred to as vision-aided map matching on parallel roads. It may basically use a video camera on board the vehicle, for instance, and try to localize an, for instance moving, vehicle by comparing the vision-measured road width and finding the matched width stored in the map database for the roads, thus determining the correct link on parallel roads and resolving the ambiguity the state-of-the-art navigation systems commonly faces.

In another exemplary embodiment, the method works as follows. First, it is assumed that all the check points p (Figs. 1, 2) where the vision-aided map matching, the exemplary embodiment of the inventive method, is necessary, is marked in the map database. Then, as the vehicle moves, the on-board navigation system checks the map database for upcoming parallel roads marked by check points p (Figs. 1, 2). If a check point p is identified, the navigation system calculates the distance after that check point p where the vision-aided map-matching should initiate road width measurement. Later, when the vehicle passes beyond the check point p, for instance after a certain distance between 1 Om and 20m, the vision system can start to look for curbs and/or road edges on the side on the road and calculate the lateral distance between two curbs and/or edges of the road. On flat parallel roads, it is possible that the system recognizes two pairs of paralleled curbs and/or road edges. In this case, for instance the pair dominating the center of the image shall represent the curbs and/or edges of the current road H (Figs. 1, 2) that the vehicle is currently driving on.

The measured road width wi (Figs. 1, 2) is then compared with the widths of the parallel roads that are stored in the map database. In this example, the link with a matched road width is then the current road R (Figs. 1, 2) the vehicle is currently travelling on, be it the main road or the parallel side road, and the position of the vehicle will be indicated on a display of the navigation system correctly. Thus, the traditional uncertainty issues in a navigation system for map matching on parallel roads can be solved with great confidence.

List of reference signs 1 step 2 step 3 step 4 step step 6 step M map p check point Wi width w2 width xl vehicle's position x2 vehicle's position R current road forward direction y alternative position

Claims (4)

1. Claims Method for determining a position (xl, x2, y) of a vehicle by a navigation system of the vehicle, with the navigation system -receiving of position data by aid of a positioning system, -determining of the vehicle's position (xl, x2, y) on a map (M) stored in the navigation system, characterized by the navigation system furthermore a) receiving data on a width (wi) of a current road (R) the vehicle is currently travelling on from a camera system of the vehicle, b) verifying if the vehicle's position (xl, x2, y) on the map (M) is consistent with the width (wl) of the current road (R), and c) correcting the vehicle's position (xl, x2, y) on the map (M) to a new position (x2) on the map (M) that is consistent with the width (wi) of the current road (R) if the vehicle's position (xl, x2, y) on the map it has determined before is not consistent with the width (wi) of the current road (H).
2. 2. Method according to claim 1, characterized by the map (M) stored in the navigation system having specific locations marked as check points (p), and the steps a), b), and c) being only executed if the vehicle's position (xl, x2, y) on the map (M) is close to one of the check points (p), in particular if the vehicle, according to it's position (xl, x2, y) on the map (M), is passing a check point (p) or has moved away a predefined distance from a check point (p) after passing it.
3. 3. Method according to claim 2, characterized by the specific locations being entry points of parallel roads and/or ramps.
4. 4. Method according to claim 3, characterized by -comparing of the widths (wl, w2) that the parallel roads have according to the map (M) preceding the steps a), b), and c), and the steps a), b), and c) being only executed if the parallel roads' widths (wi, w2) are different, in particular if they differ enough so that the data from the camera system is sufficiently precise for the navigation system to recognize the difference of the widths (wi, w2).