DE102017011982A1 - Method for route forecasting - Google Patents

Abstract

The invention relates to a method for distance forecasting, wherein based on vehicle sensor data and a digital map, a route in front of the vehicle (S1, S2) is determined. According to the invention, a trajectory of a distance traveled by the vehicle (1), based on the vehicle sensor data, is recorded relative to a vehicle-fixed or earth-fixed coordinate system. The recorded trajectory is imaged in the digital map using at least one statistical method in such a way that map information is present in the vehicle-fixed or earth-fixed coordinate system. Based on this figure, the route (S1, S2) located in front of the vehicle (1) is determined.

Description

The invention relates to a method for forward planning according to the preamble of claim 1.

From the DE 10 2006 040 334 A1 For example, there is known a method for track detection with a driver assistance system of a vehicle comprising a sensor system for lane detection. With the sensor system for lane detection lane markings are detected in a lying in front of the vehicle area of a traffic lane, where lane markers are assigned to nodes with coordinates of a first coordinate system. The coordinates of the interpolation points are converted into a second coordinate system, the course of lane markings and lanes being reconstructed from the position of the interpolation points in the second coordinate system.

The invention is based on the object to provide a comparison with the prior art improved method for distance forecasting.

The object is achieved by a method having the features specified in claim 1.

In the method for route forecasting, a route in front of the vehicle, for example road and / or lane course, is determined on the basis of vehicle sensor data and a digital map.

According to the invention, based on the vehicle sensor data, a trajectory of a distance traveled by the vehicle is recorded with reference to a vehicle-fixed or earth-fixed coordinate system. The recorded trajectory is imaged in the digital map using at least one statistical method in such a way that map information is present in the vehicle-fixed or earth-fixed coordinate system. On the basis of this illustration of the trajectory in the digital map, the route ahead of the vehicle is determined.

The method enables highly accurate fitting of a high accuracy digital map into on-board information, i. H. in the vehicle-fixed or earth-fixed coordinate system, by bringing the vehicle sensor data in relation to the digital map. The digital card can thus replace a lane detection sensor and is characterized by very large detection ranges with high accuracy. Objects at a great distance can be assigned to a corresponding lane. Thus, it is possible, for example, at track constrictions, d. H. in case of a track overlap of an object to make a brake or avoidance decision.

Thus, the method allows in particular in a highly automated driving operation of a vehicle a route forecast with a long range and high accuracy and consequently an improved highly automated driving operation.

Embodiments of the invention are explained in more detail below with reference to drawings.

• 1 schematisch einen Streckenverlauf einer digitalen Karte und einen mittels Fahrzeugsensordaten ermittelten Streckenverlauf in einer Ausgangssituation,
• 2 schematisch die Streckenverläufe gemäß 1 nach der Durchführung eines ersten und eines zweiten Verfahrensschritts,
• 3 schematisch die Streckenverläufe gemäß 2 nach der Durchführung eines dritten Verfahrensschritts,
• 4 schematisch Diagramme einer Plausibilitätsprüfung und einer Ausrichtung eines Streckenverlaufs einer digitalen Karte sowie eines mittels Fahrzeugsensordaten ermittelten Streckenverlaufs,
• 5 schematisch die Streckenverläufe gemäß 3 nach der Durchführung eines vierten Verfahrensschritts,
• 6 schematisch einen Betrag eines maximalen lateralen Fehlers einer digitalen Karte und
• 7 schematisch einen weiteren Streckenverlauf einer digitalen Karte und einen mittels Fahrzeugsensordaten ermittelten weiteren Streckenverlauf nach der Durchführung eines ersten und zweiten Verfahrensschritts.

Showing:

• 1 2 schematically shows a route course of a digital map and a route course determined by means of vehicle sensor data in an initial situation,
• 2 schematically the routes according to 1 after carrying out a first and a second process step,
• 3 schematically the routes according to 2 after carrying out a third process step,
• 4 2 schematically diagrams of a plausibility check and an alignment of a route of a digital map and of a route determined by means of vehicle sensor data,
• 5 schematically the routes according to 3 after the implementation of a fourth procedural step,
• 6 schematically an amount of maximum lateral error of a digital map and
• 7 schematically a further route of a digital map and a determined by means of vehicle sensor data further route after performing a first and second method step.

Corresponding parts are provided in all figures with the same reference numerals.

In 1 are a route determined by means of vehicle sensor data S1 and a route S2 a digital map presented in a starting situation.

During a so-called highly automated driving operation of a vehicle 1 Data is used by high-precision digital maps. These digital maps should provide a route preview with regard to the track course with the greatest possible range. Furthermore, by means of an on-board on-board environment sensor of a vehicle 1 Objects and a lane detected in a vehicle-fixed or Erdfesten coordinate system as vehicle sensor data. In this case, an already traveled path is also determined, for which purpose, for example, a vehicle speed and a yaw rate of the vehicle 1 to get integrated. Thus the route S2 the digital map in the vehicle 1 can be used, the digital map is transferred to the vehicle-fixed or earth-fixed coordinate system.

Using erdfester coordinate systems for tracking a vehicle position, for example, as in " Petrovskaya, Anna and Thrun, Sebastian: Model based vehicle detection and tracking for autonomous urban driving; In: Autonomous Robots Volume 26 Ed. 2-3, April 2009, pp. 123-139, Kluwer Academic Publishers Hingham, MA, USA ISSN: 0929-5593 "And allows for a preview in a terrestrial coordinate system.

In the illustrated starting situation, the route is S2 the digital map and the route determined by means of the vehicle sensor data S1 shifted to each other. To the route S2 the digital map with the determined by means of the vehicle sensor data route S1 to combine such that with a long range of the route forecast and at the same time a high accuracy, ie a high degree of agreement of a real route S2 the digital map with the determined route S1 is achieved, based on the vehicle sensor data is a trajectory of the vehicle 1 already traveled distance recorded based on the vehicle-fixed or ground-fixed coordinate system. Subsequently, the recorded trajectory is imaged in the digital map using at least one statistical method such that map information is present in the vehicle-fixed or ground-fixed coordinate system. Based on this illustration, the front of the vehicle 1 located road and / or lane course determined.

For this purpose, the two routes S1 . S2 according to 2 in a first method step in such a way shifted to each other that this at a current vehicle position P congruent superimposed. This approximation of the routes S1 . S2 in particular takes place about a position of a rightmost lane marking a first lane. Error calculations are made, for example, for each point in the data set, for example at intervals of one meter each. This results in an accuracy of the approximation and thus the route forecast from an accuracy of a localization of the vehicle 1 , an accuracy of the alignment of the routes S1 . S2 and an accuracy of the digital map.

Subsequently, in a second process step on each route S1 . S2 one point each Q1 . Q2 determined by starting from the vehicle position P a given distance on the respective route S1 . S2 is recalculated. The distance is in a possible embodiment 200 meters.

The two points Q1 . Q2 each with a straight line with the vehicle position P connected so that two secants SK1 . SK2 form. Furthermore, between the secants SK1 . SK2 an angle β determined.

As in 3 are displayed, then the route courses S1 . S2 such to the vehicle position P turned that angle β Becomes zero. This creates a distance L starting from the vehicle position P a deviation .DELTA.w ,

4 shows diagrams D1 to D3 of a plausibility check of a route S2 a digital map and a route determined by means of vehicle sensor data S1 , wherein the same aligned data records, ie route courses S1 . S2 , according to the descriptions too 2 and 3 moved and around the vehicle position P around the angle β were filmed.

Further shows 4 a diagram D4 of a result of a pre-equalization of the routes S1 . S2 , a diagram D5 of a result of a translational alignment of the courses S1 . S2 and a diagram D6 of a result of approximation of secants SK1 . SK2 , In this case, an alignment of the position of the rightmost lane marking of the first lane by means of the same method, whereby a final error at a predetermined distance is determined as Euclidean distance.

In 5 are the routes S1 . S2 according to 3 after performing a fourth method step, assuming that is an absolute lateral error of the route S2 the digital card is limited. This causes a relative lateral error of the route S2 the digital map is also limited and less than or equal to the absolute lateral error. An approximation of the routes S1 . S2 according to 5 takes place using the method of least squares.

An amount A the maximum lateral error of the route S2 the digital map is in 6 shown in more detail.

eine Genauigkeit bzw. Exaktheit an jedem Punkt des Streckenverlaufs S2 kleiner als der Betrag A des maximalen lateralen Fehlers des Streckenverlaufs S2 der digitalen Karte oder gleich diesem Betrag A ist.For the optimized alignment of both routes S1 . S2 according to the method of least squares, it is assumed that according to $$\left|\mathrm{\Delta }w\text{'}\right|\le A$$

an accuracy at each point of the route S2 less than the amount A the maximum lateral error of the route S2 the digital map or equal to this amount A is.

It follows that the alignment of the routes S1 . S2 at the vehicle position P an error with the amount A introduces. This also applies to the points Q1 . Q2 , It follows that a lateral alignment of the courses S1 . S2 a mistake with twice the amount A having.

so dass $$\left|\mathrm{\Delta }w\prime \right|=\text{tan}\beta =L\cdot \frac{2A}{D}$$

ist.For the secants SK1 . SK2 applies: $$\text{tan}\beta =\frac{2A}{D}.$$

so that $$\left|\mathrm{\Delta }w\text{'}\right|=\text{tan}\beta =L\cdot \frac{2A}{D}$$

is.

It follows $$\left|\mathrm{\Delta }w\right|\le A+A+L\cdot \frac{2A}{D}\text{,}$$

Here, the first A in equation (4) represents the relative accuracy, that is, the amount A the relative error, and that second A is the amount A the error in the alignment of the two routes S1 . S2 at the vehicle position P The third term of equation (4) describes the rotation about the vehicle position P ie the angle β ,

It follows $$\left|\mathrm{\Delta }w\right|\le 2A\left(1+\frac{L}{D}\right)\text{,}$$

In 7 are another route S2 a digital map and another determined by means of vehicle sensor data further route S1 after performing the first and second process steps, ie after the shift to coincidence at the current vehicle position P and the determination of the secants SK1 . SK2 represented. Here is the angle β between the secants SK1 . SK2 equals zero.

ergibt.It follows that no rotation of the routes S1 . S2 is done to each other, so that the amount of deviation .DELTA.w according to $$\left|\mathrm{\Delta }w\right|\le A+L\cdot \text{tan}\gamma$$

results.

In this case, y is an orientation error between a direction of travel determined according to data of the digital direction of travel and a direction of travel determined from ground data.

LIST OF REFERENCE NUMBERS

1

vehicle

A

amount

D1 to D6

diagram

L

distance

P

vehicle position

Q1

Point

Q2

Point

S1

route

S2

route

SK1

secant

SK2

secant

β

angle

.DELTA.w

deviation

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 102006040334 A1 [0002]

Cited non-patent literature

• Petrovskaya, Anna and Thrun, Sebastian: Model based vehicle detection and tracking for autonomous urban driving; In: Autonomous Robots Volume 26 Ed. 2-3, April 2009, pp. 123-139, Kluwer Academic Publishers Hingham, MA, USA ISSN: 0929-5593 [0014]

Claims (1)

Method for distance forecasting, wherein on the basis of vehicle sensor data and a digital map located in front of the vehicle (1) route (S1, S2) is determined, characterized in that - based on the vehicle sensor data a trajectory of the vehicle (1) already traveled distance based on a vehicle-fixed or earth-fixed coordinate system is recorded, the recorded trajectory is imaged in the digital map using at least one statistical method in such a way that map information is present in the vehicle-fixed or ground-based coordinate system, and based on the map of the route ahead of the vehicle (FIG. S1, S2) is determined.

Images