Method for modeling a motor vehicle sensor in a virtual test environment

Abstract

The disclosure relates to a method that models a motor vehicle sensor in a virtual test environment by way of definition. Using a sensor support, a raycast distribution shape, a group of raycast properties, a raycast reflection factor, and a raycast echo, a sensor in reality may be tested in a virtual environment to calibrate the sensor in reality. The sensor support is a virtual sensor support for a virtual sensor model, which forms a three-dimensional or two-dimensional avatar of the sensor in reality, in the virtual test environment. The sensor support has a sensor starting point that is used as an origin for a raycast distribution shape. The method extracts a special application of the sensor in reality in an application case, which is particularly useful for testing scenarios.

Claims

1. A method that models a motor vehicle sensor comprising: obtaining sensor model parameters from a plurality of physical sensor parameters to model the sensor in a virtual test environment by way of definition and use of a virtual sensor model including: a sensor support for the sensor in the virtual test environment, and includes a sensor starting surface that is used as an origin for a raycast distribution shape, wherein the raycast distribution shape is a predefined two-dimensional or three-dimensional shape, an origin of which is a starting surface of the sensor support, wherein the raycast distribution shape has a plurality of evenly distributed raycasts, an origin of which is either the starting surface of the sensor support or a point in a starting plane of the sensor support; wherein the group of raycast properties are a damping, a propagation speed, and a detection accuracy of raycasts, wherein the damping is the same for all raycasts use either the same predefined value or are determined using the same method and is defined as a single value that specifies a percentage of original signal that would return if an object were struck by the raycast at a specific distance from an origin, wherein the propagation speed is the same for all raycasts use either the same predefined value or are determined using the same method and defines a delay, after which, at a distance at which a raycast strikes an object, an echo value of the raycast is available in the sensor, and wherein the detection accuracy is identical for all raycasts use either the same predefined value or are determined using the same method and is a probability that a raycast sends back an incorrect value, wherein the raycast reflection factor is associated with objects present in the virtual test environment and is a percentage that indicates how well an object reflects waves emitted by the sensor, and wherein the raycast echo is a collection of coordinates and echo values sent back by each raycast, wherein the echo values are dependent on the detection accuracy, the damping, the raycast reflection factor, and whether a hit is present.

2. The method as claimed in claim 1, wherein the sensor is a sensor operative based on electromagnetic or acoustic waves.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 shows an illustration of a raycast according the disclosure;

(2) FIG. 2 shows a system overview of an interaction between parts of a sensor model according the disclosure; and

(3) FIG. 3 shows an example of a sensor model according to an embodiment of the disclosure.

DETAILED DESCRIPTION

(4) As required, detailed embodiments of the present disclosure are disclosed herein; however, it is to be understood that the disclosed embodiments are merely exemplary of the disclosure that may be embodied in various and alternative forms. The figures are not necessarily to scale; some features may be exaggerated or minimized to show details of particular components. Therefore, specific structural and functional details disclosed herein are not to be interpreted as limiting, but merely as a representative basis for teaching one skilled in the art to variously employ the present disclosure.

(5) To explain raycasts, reference will firstly be made to FIG. 1.

(6) In a computer game (or in a computer game development environment, a game engine), a raycast or “vision beam” is something represented by at least one vector D in three dimensions, in a three-dimensional reference coordinate system xyz, which has a specific length (up to infinite) and an application point or origin C placed on an object A. Raycasts offer a possibility of finding all objects B that are located on a trajectory thereof, which in particular enables coordinates of contact points E between the raycasts and these objects to be found. Raycasts are typically used to emulate instantaneous projectile trajectories, to ascertain visibility of objects, etc.

(7) The sensor model consists of the following parts illustrated in FIG. 2:

(8) A definition of a sensor support 1.

(9) The sensor support 1 is an “avatar” of a sensor in a virtual test environment and has the following properties:

(10) a starting point or a starting surface (i.e., an origin) for a raycast distribution shape 2;

(11) a 3D rotational movement as a function of time, which indicates whether the sensor can, for example, rotate at a specific angular velocity or whether the sensor is fixed.

(12) A raycast distribution shape 2:

(13) The raycast distribution shape 2 is a three-dimensional shape (for example, a cone, a cylinder, or a conventional volume), an origin of which is the starting point or the starting surface of the sensor support 1.

(14) Dimensions of the raycast distribution shape 2 model a maximum range of the sensor in three-dimensional space.

(15) The raycast distribution contains N raycasts, which are evenly distributed in the raycast distribution shape 2. The origin of each raycast is either the starting point of the sensor support 1 or a point in a starting plane of the sensor support 1.

(16) The number of raycasts results in a spatial resolution of the sensor. They can also be used to model discretization of items of sensor information.

(17) A definition of a raycast damping 3:

(18) The raycast damping 3 can be identical for all raycasts of the raycast distribution shape 2 or can be specific for each raycast.

(19) The damping 3 can be a coefficient or a formula within the raycast distribution shape 2, which specifies a percentage of original signal that would return if an object were struck by the raycast at a specific distance from the origin.

(20) A definition of a raycast propagation speed 4:

(21) The propagation speed 4 can be identical for all raycasts of the raycast distribution shape 2 or can be specific for each raycast.

(22) The propagation speed 4 is used to define a delay, after which, at a distance at which the raycast strikes an object, an echo value of the raycast is available in the sensor.

(23) The propagation speed 4 can be used for modeling a nature of waves emitted by the sensor (sound, light, electromagnetic, etc.) and of environmental influences, for example, weather, temperature, etc.

(24) A definition of a raycast detection accuracy 5:

(25) The detection accuracy 5 can be identical for all raycasts of the raycast distribution shape 2 or can be specific for each raycast.

(26) The detection accuracy 5 is a correction coefficient that enables an error in the echo value of the raycast to be emulated. It is a probability that a raycast sends back an incorrect value (for example, primarily because of a lack of target, possible false alarm).

(27) A definition of a reflection factor 6:

(28) The definition of a reflection factor 6 is associated with objects present in the virtual test environment.

(29) The reflection factor is a percentage that represents how well an object reflects the wave emitted by the sensor.

(30) The reflection factor enables materials to be modeled, which only reflect weakly because of a shape, structure, nature, etc. thereof, for example, some items of clothing.

(31) A definition of a raycast echo 7:

(32) Each raycast can send back a collection of values, which are called an echo, namely:

(33) a 3D coordinate, which corresponds to a point of incidence with an object (or not, if there are no hits), and

(34) an echo value lying in a range [0, . . . , 1], which is dependent on the detection accuracy 5, the value of the damping 3, the reflection factor 6 of the object struck, and on whether a hit is present or not. It can be provided by the following equation:
accuracy value=(detection accuracy)×(damping)×(reflection factor)×(1 for a hit or otherwise 0)

(35) Each value of a raycast echo 7 is available after a propagation time, which is determined by the propagation speed 4 defined for this raycast and a distance of the struck object from the origin of the raycast.

(36) If necessary, it is also possible to model a sensor in only two dimensions.

(37) FIG. 3 shows an example of a sensor model. In FIG. 3, G is the sensor support 1, which consists of a cylinder and is a 3D avatar of the sensor, which is placed in a virtual environment and on a starting point H, which is an origin of the raycast distribution shape M. The raycast distribution shape M is a cone; the dimensions of the cone supply a maximum range of the sensor in three-dimensional space. This raycast distribution shape M contains four raycasts R, which means that this sensor can detect an object that traverses these raycasts R. For each raycast R, a damping 3, a propagation speed 4, and a detection accuracy 5 can be defined.

(38) It is possible to generate the sensor model automatically, inter alia, based on the following:

(39) Measurements of the sensor in reality that are analyzed by an algorithm for machine learning, which is trained to extract parameters required for modeling the sensor.

(40) Test cases or application cases that are specified in a machine-interpretable language (for example, a pseudo-natural language) and analyzed by an algorithm, which extracts a special application of the sensor in the application case. This is particularly useful for testing error scenarios or safety-related scenarios.

(41) Sensor data sheets, which are available in a standard format, for example, XML, and can be provided by a sensor supplier. An algorithm, for example, a serializer, can extract the parameters required for modeling the sensor from such digital data sheets.

(42) In summary, the following methods and/or algorithms are essential:

(43) A method for modeling a virtual motor vehicle sensor in a three-dimensional virtual environment by way of definition and use of a sensor support 1, a raycast distribution shape 2, raycast properties, in particular its damping 3, propagation speed 4, and detection accuracy 5, a raycast reflection factor 6, and a raycast echo 7, which is a returned value of the raycast.

(44) A method for defining a virtual sensor support for a virtual sensor model, having

(45) a three-dimensional or two-dimensional avatar of the sensor in a solely graphical representation;

(46) a sensor starting point or a sensor starting surface that is used as an origin for a raycast distribution shape or its raycasts; and

(47) a definition of a sensor support movement, which describes whether the sensor support is a fixed support or whether the sensor support can move after attachment to a virtual object (for example, as is the case with lidar or rotating radar). If the sensor can move, the associated movement is also described (for example, rotational velocity, etc.).

(48) A method for defining a raycast distribution shape for a virtual sensor model, wherein:

(49) the raycast distribution shape is a three-dimensional (e.g., a cone, a cylinder, or a conventional volume) or two-dimensional shape, an origin of which is the starting point or the starting surface of the sensor support;

(50) the dimensions of the raycast distribution shape model the maximum range of the sensor in three-dimensional or two-dimensional space;

(51) the raycast distribution contains N raycasts, which follow a specific distribution in the raycast distribution shape, typically an even distribution, wherein an origin of each raycast is either the starting point of the sensor support or a point in a starting plane of the sensor support; and

(52) the number of the raycasts results in a spatial resolution of the sensor and can also be used to model discretization of items of sensor information.

(53) A method for defining a damping for each raycast of a raycast distribution shape, wherein:

(54) the damping is the same for all raycasts of the raycast distribution shape or is specific for each raycast, in order to model boundaries or limit conditions;

(55) the damping is typically expressed as a percentage in the range [0, . . . , 1]; and

(56) the damping is defined as a single value or a formula, which specifies a percentage of original signal that would return if an object were struck at a specific distance from the origin by the raycast within the raycast distribution shape.

(57) A method for defining a raycast propagation speed, wherein:

(58) the propagation speed is the same for all raycasts of the raycast distribution shape or is specific for each raycast;

(59) the propagation speed is used to define a delay after which, at a distance at which a raycast strikes an object, an echo value of the raycast is available in the sensor; and

(60) the propagation speed is used to model the nature of the waves (sound, light, electromagnetic, etc.) emitted by the sensor and environmental influences, for example, weather, temperature, etc.

(61) A method for defining a raycast detection accuracy, wherein:

(62) a detection accuracy is identical for all raycasts of the raycast distribution shape or is specific for each raycast; and

(63) the detection accuracy is a correction coefficient that enables an error in the echo value of the raycast to be emulated. It is a probability that a raycast sends back an incorrect value (for example, primarily because of a lack of target, possible false alarm).

(64) A method for defining a raycast detection accuracy, wherein:

(65) a detection accuracy is the same for all raycasts of the raycast distribution shape or is specific for each raycast;

(66) the detection accuracy is typically expressed as a percentage in the range [0, . . . , 1];

(67) the detection accuracy is a correction coefficient that enables an error in the echo value of the raycast to be emulated (for example, it can represent a probability that a raycast sends back an incorrect value (for example, primarily because of a lack of target, possible false alarm)); and

(68) the detection accuracy is used in an early phase of a project to emulate errors that are induced by signal processing algorithms embedded in a sensor.

(69) A method for defining an activation time pattern for each raycast of a raycast distribution shape, wherein the activation time pattern is used to specify when each raycast of a sensor is activated (e.g., all together, one after another with a specific delay, etc.).

(70) A method for defining a reflection factor for a virtual object of a virtual sensor test, wherein:

(71) a definition of a reflection factor is associated with a specific material, which is associated with each object in the virtual test environment;

(72) the reflection factor is typically expressed as a percentage in the range [0, . . . , 1];

(73) the reflection factor is a percentage that represents how well an object reflects waves emitted by the sensor; and

(74) the reflection factor enables materials to be modeled that only reflect weakly because of the shape, structure, nature, etc. thereof (for example, some pieces of clothing).

(75) A method for defining an echo for each raycast of a sensor model, wherein:

(76) an echo is processed for each raycast of the sensor model;

(77) an echo is a value in the range [0, . . . , 1], which is a function of the damping value, the detection accuracy, and the reflection factor of the object struck by the raycast and of the presence of a hit;

(78) an echo value is available after a time delay, which is a function of the raycast propagation speed and a distance of a struck object; and

(79) an echo value is associated with three-dimensional or two-dimensional coordinates of the struck object.

(80) A method for automatically generating a virtual sensor model, wherein:

(81) the method is an algorithm for machine learning, which is trained to extract parameters required that model the virtual sensor from measurements of the sensor in reality;

(82) the method is an algorithm, which analyzes a test case or application case that is specified in a machine-interpretable language (for example, a pseudo-natural language), and extracts a special application of the sensor in a test case or application case (particularly useful for testing error scenarios or safety-related scenarios); and

(83) the method is an algorithm, which extracts parameters required for modeling the sensor from a sensor data sheet, which is available in a standard format (for example, XML) and is provided by a sensor supplier.

(84) While exemplary embodiments are described above, it is not intended that these embodiments describe all possible forms of the disclosure. Rather, the words used in the specification are words of description rather than limitation, and it is understood that various changes may be made without departing from the spirit and scope of the disclosure. Additionally, the features of various implementing embodiments may be combined to form further embodiments of the disclosure.