DE102019209081A1 - Correction method, correction device and vehicle - Google Patents

Abstract

The present invention generally relates to a correction method for a first and a second sensor coordinate set, which are determined with a distance sensor (202) which is attached to a vehicle (200), comprising:
Determining (101) the first sensor coordinate set for a first measurement time;
Determining (102) the second set of sensor coordinates for a second measurement time, the second set of sensor coordinates being able to be represented by shifting and rotating the first set of sensor coordinates;
Determining (103) a data set which has comparison values for the first and second sensor coordinate sets; and
Compensating (104) the rotation on the basis of a coordinate transformation of the comparison values, whereby the second sensor coordinate set is corrected.

Description

The present invention relates generally to a correction method, a correction device and a vehicle.

In general, various methods are known to determine or correct an angular misalignment of a distance sensor (for example lidar or radar) which is attached to a vehicle, such as, for example, various image processing methods which assign detected point clouds to a grid; Combinations of 2D and 3D lidar; designated landmarks; multiple sensors; Geometry; Real-time lidar odometry and mapping (LOAM) and the like.

In the event of an angular misalignment of the distance sensor, a measured point or a measured distance can be incorrectly determined, which can result in inaccuracies, for example in road traffic or also in (self-propelled) agricultural machinery or warehouse vehicles.

Even if solutions for correcting an angular misalignment of a distance sensor are known from the prior art, it is an object of the present invention to provide a correction method, a correction device and a vehicle.

The object is achieved by the correction method according to claim 1, the correction device according to claim 5 and the vehicle according to claim 6.

• Ermitteln eines ersten Sensor-Koordinatensatzes für einen ersten Messzeitpunkt;
• Ermitteln eines zweiten Sensor-Koordinatensatzes für einen zweiten Messzeitpunkt, wobei der zweite Sensor-Koordinatensatz durch eine Verschiebung und durch eine Rotation des ersten Sensor-Koordinatensatzes darstellbar ist;
• Ermitteln eines Datensatzes, welcher Vergleichswerte für den ersten und den zweiten Sensor-Koordinatensatz aufweist; und
• Ausgleichen der Rotation anhand einer Koordinatentransformation der Vergleichswerte, wodurch der zweite Sensor-Koordinatensatz korrigiert wird.

According to a first aspect, the present invention provides a correction method comprising:

• Determining a first sensor coordinate set for a first measurement time;
• Determining a second set of sensor coordinates for a second measurement time, the second set of sensor coordinates being able to be represented by a shift and a rotation of the first set of sensor coordinates;
• Determining a data record which has comparison values for the first and second sensor coordinate sets; and
• Compensating for the rotation using a coordinate transformation of the comparison values, whereby the second sensor coordinate set is corrected.

According to a second aspect, the present invention provides a correction device which is set up to carry out the correction method according to the first aspect.

According to a third aspect, the present invention provides a vehicle comprising a correction device according to the second aspect.

Further advantageous embodiments of the invention emerge from the dependent claims, the drawings and the following description of preferred exemplary embodiments.

• Ermitteln des ersten Sensor-Koordinatensatzes für einen ersten Messzeitpunkt;
• Ermitteln des zweiten Sensor-Koordinatensatzes für einen zweiten Messzeitpunkt, wobei der zweite Sensor-Koordinatensatz durch eine Verschiebung und durch eine Rotation des ersten Sensor-Koordinatensatzes darstellbar ist;
• Ermitteln eines Datensatzes, welcher Vergleichswerte für den ersten und den zweiten Sensor-Koordinatensatz aufweist; und
• Ausgleichen der Rotation anhand einer Koordinatentransformation der Vergleichswerte, wodurch der zweite Sensor-Koordinatensatz korrigiert wird.

Some exemplary embodiments relate to a correction method for a first and a second sensor coordinate set, which are determined using a distance sensor that is attached to a vehicle, comprising:

• Determining the first sensor coordinate set for a first measurement time;
• Determining the second set of sensor coordinates for a second measurement time, the second set of sensor coordinates being able to be represented by shifting and rotating the first set of sensor coordinates;
• Determining a data record which has comparison values for the first and second sensor coordinate sets; and
• Compensating for the rotation using a coordinate transformation of the comparison values, whereby the second sensor coordinate set is corrected.

The correction method can generally be carried out by a processor (for example a CPU (Central Processing Unit)), computer, server, control unit in a vehicle or the like, or for example with a correction device according to the second aspect.

In some exemplary embodiments, the first sensor coordinate set is determined with a distance sensor (e.g. lidar or radar) which transfers points (location points) into a coordinate system or the measured distances can be represented as coordinates (or as a coordinate set) .

In addition, the points can be displayed as a two-dimensional (2D) or three-dimensional (3D) point cloud, which has the advantage of visualization, or as a pure data set, wherein the data set can advantageously be processed.

In some exemplary embodiments, the distance sensor is permanently attached to a vehicle so that distances to objects can be detected and the distance sensor typically has a fixed orientation to the vehicle (which is however, it can change, for example, through material deformation, temperature influences, etc.).

The distance information can be used for autonomous movement of the vehicle, to maintain a predetermined distance to an object (e.g. a preceding vehicle or an object with a fixed position) or the like, the objects being recorded, for example, as point clouds with the aid of one of the sensor coordinate sets can.

The vehicle can generally be any type of vehicle, for example passenger cars, trucks, agricultural machines, tractors, industrial trucks (e.g. lifting trucks, forklifts) that have a distance sensor according to the invention.

Furthermore, as is generally known, various individual measurements (which are organized, for example, in frames) can be recorded. A first individual measurement can be used to determine the first sensor coordinate set for a first measurement time and a second individual measurement can be used to determine the second sensor coordinate set for a second measurement time (the second measurement time can be after the first measurement time, without loss of generality ).

In some exemplary embodiments, determining includes measuring, calculating, or the like of the sensor coordinate sets. Known methods or known sensors can be used for this, which has the advantage of saving time.

In some exemplary embodiments, a measurement time relates to a time at which a distance or position determination is carried out or coordinates or points for a point cloud are determined.

The first and second measurement times are not necessarily successive times at which individual measurements are carried out. However, this case is not excluded either. For example, the second set of sensor coordinates can be determined after a predetermined (or random) number of individual measurements.

The advantage of this in some exemplary embodiments is that a disturbance can be compensated because an angular misalignment of the distance sensor typically cannot be determined directly or not alone in the immediately following individual measurement, but only after several individual measurements (e.g. because of a readjustment of the distance sensor, and the same).

In some exemplary embodiments, the second set of sensor coordinates can be represented by a (mathematical) shift and by a rotation of the first set of sensor coordinates.

In general, a shift and / or a rotation of the sensor coordinate records can be caused by a shift and / or a rotation of the distance sensor.

For example, in embodiments in which a sensor is mounted in a vehicle, the vehicle (and thus the sensor) can move between the first and second points in time, which is generally achieved by a translation (displacement) and a rotation of coordinates (ie the first sensor coordinate set) can be displayed.

As a result, the movement of the first sensor coordinate set can advantageously be illustrated and used for calculations (or data processing) in order to arrive at the second sensor coordinate set.

Furthermore, in some exemplary embodiments, a data record is determined which has comparison values for the first and second sensor coordinate sets.

Like the first and second sensor coordinate sets, the data record can be based on sensor data that can be represented by coordinates. The sensor data can be recorded by the same sensor that records the first and second sensor coordinate sets, or by a different sensor.

For example, the data record can be based on inertia sensor data which determine a position, an orientation, a speed, a direction, an acceleration, or the like.

Several sensors can also be used to determine the data record, e.g. B. several inertial sensors and / or video sensors (z. B. cameras) and / or further distance sensors, or the like, which are set up (also in combination), a position, orientation, location, speed, direction, acceleration, or the like determine.

An angular misalignment of the distance sensor can thus be recognized on the basis of the data record, which is used as a reference and thus supplies comparison values.

Thus, in some exemplary embodiments, the rotation of the second sensor coordinate set (or the second sensor coordinate system) is compared to the first sensor coordinate set (or to the first sensor coordinate system) balanced.

As a result, it is advantageously not necessary to compensate for an angular misalignment of the distance sensor by means of a new adjustment, which is time-consuming and costly, but can be corrected by the method according to the invention.

In some exemplary embodiments, the coordinate transformation includes a comparison or a calculation which compares the data set, represented as coordinates, with the first and / or the second sensor coordinate set and transfers it into a reference system of the first and / or second sensor coordinate set, as it did is well known.

As a result, known mathematical formalisms can be used to convert the data record and the sensor coordinate records into one another.

The correction (“correct”) according to the invention can therefore readjust the second sensor coordinate set, which arises from an undesired rotation of a sensor after the first sensor coordinate set has been recorded, so that, as discussed, no physical readjustment (or position or Alignment correction) of the sensor is required.

In some exemplary embodiments, the first and second sets of coordinates are compared using an iterative nearest-point algorithm.

Iterative closest point algorithms (or iterative closest point algorithm, ICP for short) are generally known to those skilled in the art.

In some exemplary embodiments, the ICP can be used to record a translation and a rotation of point clouds (or sets of coordinates which can also be present as pure data or of coordinate systems, such as the sensor coordinate systems).

For example, in some exemplary embodiments, the first and the second set of coordinates are compared by determining a translation and a rotation of the second set of coordinates with respect to the first set of coordinates.

In this way a movement (i.e. translation and / or rotation) of the distance sensor can be detected.

In some exemplary embodiments, the data record is determined by means of egomotion estimation.

In general, egomotion estimation (or egomotion estimation) is known to the person skilled in the art, which offers the advantage that a known method can be used according to the invention.

In the egomotion estimation, in some exemplary embodiments a position, direction, orientation, speed, acceleration, and the like are determined on the basis of further sensor data which correspond to the data set according to the invention. The additional sensor data can also be determined by multiple sensors, as discussed herein.

The further sensor data (hereinafter: Egomotion data) are used in some exemplary embodiments to estimate the position of the vehicle. This position can be represented by an Egomotion coordinate system or a point cloud which has comparison values for the first set of coordinates at the first measurement time and comparison values for the second set of coordinates at the second measurement time.

In some exemplary embodiments, the position of the vehicle is determined in a first egomotion frame (egomotion individual measurement) at the first measurement time and the position of the vehicle is determined in a second egomotion frame at the second measurement time.

In some exemplary embodiments, the Egomotion coordinate system experiences a translation and / or a rotation at the second measurement time compared to the first measurement time, just like the first and the second coordinate set (without restricting the present invention to them).

As a result, in some exemplary embodiments, the movement of the vehicle can be represented by four coordinate systems which can be converted into one another.

1 illustrates these coordinate systems in one system 1 .

The system 1 points a point 2 which part of a point cloud is and is determined with the aid of a distance sensor according to the invention.

Furthermore, the system has 1 a first distance sensor coordinate system 3 , which represents a reference system of the distance sensor at the first measurement time; a second distance sensor coordinate system 4th , which represents a reference system of the distance sensor at the second measurement time; a first egomotion Coordinate system 5 , which represents a reference system of the egomotion estimation at the first measurement time; and a second Egomotion coordinate system 6th , which represents a reference system of the Egomotion estimation at the second measurement time.

Furthermore, in the system 1 several vectors shown which translations of the coordinate systems 3 to 6th represent. In addition, these represent the point with respect to the first distance sensor coordinate system 3 and with respect to the second distance sensor coordinate system 4th represent.

The reference symbols of the vectors identify the translation of the coordinate systems or the coordinates of the point, so that, for example, the vector 53 the translation of the first Egomotion coordinate system 5 on the first distance sensor coordinate system 3 represents.

Furthermore, for example, the vector 32 the point 2 with respect to the first distance sensor coordinate system 3 dar and the vector 34 represents the translation of the first distance coordinate system 3 to the second distance coordinate system 4th dar, and so on.

The vector 34 illustrates, as mentioned, the translation of the distance sensor between the first and the second measurement time. A vectorial distance between the two distance sensor coordinate systems 3 and 4th , is referred to below as Δx _s .

A vectorial distance between the two Egomotion coordinate systems 5 and 6th (i.e. the vector 56 ) is referred to below as Δx _v .
A vectorial distance between the first Egomotion coordinate system 3 and the point 2 (i.e. the vector 32 ) is referred to below as x and the vector 42 as x '. The vector 53 is referred to below as p _s and the vector 64 as ps'.

The following calculation uses the notation introduced above:

wobei R^T eine Rotation des zweiten Sensorkoordinatensystems 4 bezüglich des ersten Sensorkoordinatensystem 3 abbildet und damit allgemein eine Rotationsmatrix ist.First is the vector 42 calculated, ie: $$\text{x '}={\text{R.}}^T\left({\text{x}}^{\text{s}}-\Delta {\text{x}}_{\text{s}}{}^{\text{s}}\right)$$

where R ^{T is} a rotation of the second sensor coordinate system 4th with respect to the first sensor coordinate system 3 and is therefore generally a rotation matrix.

Furthermore, indices above (in equation (1) a superscript s, later also a superscript e) indicate that one is in the sensor reference system (s) or in the egomotion reference system (e).

wobei die Matrix R die Rückabbildung zu R^T ist. In Gleichung (2) dient sie dazu den Vektor 64 (also p_s') durch den Vektor p_s darzustellen, und wobei die Matrix R_s ^T eine Kalibrations.- bzw. Abbildungsmatrix zwischen dem Egomotionbezugssystem (bzw. Egomotion-Koordinatensystem 5 und 6) und dem jeweiligen Sensorbezugssystem (Sensor-Koordinatensystem 3 und 4) darstellt.Multiplied and Δx _s with the vectors 53 , 56 and 64 expressed, equation (1) gives: $$\text{x}\text{'}={\text{R.}}^{\text{T}}{\text{x}}^{\text{s}}-{\text{R.}}^{\text{T}}\left(-p_{\text{s}}{}^{\text{s}}+\Delta x_v{}^s+{\mathrm{R.}}_{\text{s}}{}^T\mathrm{R.}{p}_{\text{s}}{}^e\right)$$

where the matrix R is the mapping back to R ^T. In equation (2) it is used for the vector 64 (ie p _s ') to be represented by the vector p _s , and where the matrix R _s ^{T is} a calibration or mapping matrix between the egomotion reference system (or egomotion coordinate system 5 and 6th ) and the respective sensor reference system (sensor coordinate system 3 and 4th ) represents.

Equation (2) can be rewritten as: $$\text{x '}={\text{R.}}^{\text{T}}{\text{x}}^{\text{s}}-{\text{R.}}^{\text{T}}{\text{R.}}_{\text{s}}{}^{\text{T}}\left(-{\text{p}}_{\text{s}}{}^{\text{e}}+\Delta {\text{x}}_{\text{v}}{}^{\text{e}}+{\text{Rp}}_{\text{s}}{}^{\text{e}}\right)$$

Thus the following expression can be found: $$\text{x}-\text{Rx '}={\text{R.}}_{\text{s}}{}^{\text{T}}\left(-{\text{p}}_{\text{s}}{}^{\text{e}}+\Delta {\text{x}}_{\text{v}}{}^{\text{e}}+{\text{Rp}}_{\text{s}}{}^{\text{e}}\right)$$

This allows the rotated vector 42 with respect to the vector 32 (ie overall the vector 34 ) can be represented by a rotation, as it is determined in the egomotion estimation (rotation R _s ^T ).

This can cause an angular misalignment of the distance sensor (which the two distance sensor coordinate systems 3 and 4th supplies) can be determined by the Egomotion estimate.

This angular misalignment can also be quantified with the aid of one of the rotation matrices, so that it can be compensated or corrected.

In some exemplary embodiments, the distance sensor has at least one of lidar and radar, as already discussed herein.

Some exemplary embodiments relate to a correction device which is set up to carry out the method according to the invention.

The correction device can generally be or have a processor (for example a central processing unit CPU, a graphic one Processing unit GPU and the like). Furthermore, the correction device can be or have a central on-board computer, control device or the like of a vehicle or be or have a computer which is connected to the vehicle (e.g. during a calibration process) or the like, which the first and the second sensor Coordinate set and the data set are determined and compensation data is made available, which is stored in the vehicle, or the like.

Some exemplary embodiments relate to a vehicle comprising: at least one distance sensor; and a correction device according to the invention as described herein.

• 1 ein Ausführungsbeispiel von Koordinatensystemen veranschaulicht;
• 2 ein Flussdiagramm eines erfindungsgemäßen Verfahrens ist; und
• 3 ein Blockdiagram eines Ausführungsbeispiels eines Fahrzeugs mit Korrekturvorrichtung darstellt.

Embodiments of the invention will now be described by way of example and with reference to the accompanying drawing, in which:

• 1 illustrates an embodiment of coordinate systems;
• 2 Figure 3 is a flow diagram of a method according to the invention; and
• 3 shows a block diagram of an exemplary embodiment of a vehicle with a correction device.

The following is an exemplary embodiment of a correction method 100 with reference to 2 explained which is a flow chart of the process 100 shows. As mentioned, the correction method can be used, for example, by a correction device as described herein and and further below with reference to FIGS 3 will also be described.

The procedure 100 points in 101 , determining the first sensor coordinate set, as discussed herein.

In 102 the second set of sensor coordinates is determined.

In 103 the data set is determined, for example with egomotion estimation, as discussed herein.

In 104 a determined rotation of the second sensor coordinate set is compensated on the basis of the determined data set, as described herein.

3 shows a motor vehicle 200 , which is a correction device according to the invention 201 having.

The vehicle also has a distance sensor, along with other components known to those skilled in the art (e.g. engine and the like) 202 which is designed as a lidar sensor in this exemplary embodiment, and has further sensors 203 for an egomotion estimate, with the further sensors 203 in this embodiment has a plurality of inertial sensors.

The distance sensor 202 and the other sensors 203 perform distance measurements and position determinations as described herein, whereby in each individual measurement (ie in each measurement frame and at each measurement time) a sensor coordinate set according to the invention and egomotion coordinate set (or a data set according to the invention) is generated, which is stored in a central on-board computer 204 Point clouds are generated as described herein.

The central on-board computer 204 evaluates the from the distance sensor 202 and the other sensors 203 obtained data and determines a rotation and a translation that the distance sensor 202 between two individual measurements at two measurement times. This is done with the help of an ICP algorithm based on the distance sensor 202 obtained data as described herein.

The central on-board computer also performs 204 an egomotion estimate based on the further sensors 203 obtained data from in order to generate two egomotion coordinate systems at the two measurement times and thus to determine a rotation and a translation of the two egomotion coordinate systems generated between the two measurement times. This implicitly or explicitly determines the rotation matrices explained above.

The determined coordinates and coordinate system based on the data from the distance sensor 202 and the other sensors 203 are based on a correction device according to the invention 201 transmitted which rotation due to an angular misalignment of the distance sensor 202 compensates as it is described here.

In general, several distance sensors (not shown here) can also be corrected in this way. An egomotion estimate can also be based on the data from the distance sensor 202 are carried out, whereby the further sensor system, in some exemplary embodiments, is not absolutely necessary.

List of reference symbols

1

System of coordinate systems

2

Point from point cloud

3

First distance sensor coordinate system

4th

Second distance sensor coordinate system

5

First Egomotion coordinate system

6th

Second Egomotion coordinate system

32

Vector x

34

Vector Δx _s

42

Vector x '

53

Vector p _s

56

Vector Δx _v

64

Vector p _s '

100

Method according to the invention

101

Determination of the first sensor coordinate set

102

Determination of the second set of sensor coordinates

103

Determination of data record

104

Compensate rotation

200

vehicle

201

Correction device

202

Distance sensor

203

Further sensors

204

Central on-board computer

Claims (5)

Correction method for a first and a second sensor coordinate set, which are determined with a distance sensor (202) which is attached to a vehicle (200), comprising: Determining (101) the first sensor coordinate set for a first measurement time; Determining (102) the second set of sensor coordinates for a second measurement time, the second set of sensor coordinates being able to be represented by shifting and rotating the first set of sensor coordinates; Determining (103) a data set which has comparison values for the first and second sensor coordinate sets; and Compensating (104) the rotation on the basis of a coordinate transformation of the comparison values, whereby the second sensor coordinate set is corrected. Correction procedure according to Claim 1 wherein the data set is determined by comparing the first set of coordinates and the second set of coordinates with an iterative nearest point algorithm. Correction procedure according to one of the Claims 1 or 2 , whereby the data set is determined by means of egomotion estimation. Correction device (201) which is set up to carry out the method according to one of the preceding claims. A vehicle comprising: at least one distance sensor (202); and correction device (201) according to Claim 4 .

Images