Search

CN-121995372-A - Method, computer program, device and memory medium for ascertaining the state of an object in a vehicle environment

CN121995372ACN 121995372 ACN121995372 ACN 121995372ACN-121995372-A

Abstract

The invention relates to a method (100) for ascertaining a state of an object (3) in the environment of a vehicle (1), comprising providing (101) sensor data, wherein the sensor data originate from the detection of at least one sensor (2), in particular a radar sensor, of the vehicle (1), wherein the object (3) is represented in the sensor data, ascertaining (102) at least one representation of the state using a machine learning model (50) on the basis of the provided sensor data, wherein the machine learning model (50) is trained for object detection and/or classification, ascertaining (103) at least one further representation of the state using at least one physical model on the basis of the provided sensor data. Furthermore, the invention relates to a computer program, an apparatus and a memory medium for such a purpose.

Inventors

  • F. HUTNER
  • L. SCHWARTZ
  • P. F. Rapp
  • T. Hohentanal

Assignees

  • 罗伯特·博世有限公司

Dates

Publication Date
20260508
Application Date
20251105
Priority Date
20241105

Claims (11)

  1. 1. A method (100) for determining a state of an object (3) in an environment of a vehicle (1), the method comprising: Providing (101) sensor data, wherein the sensor data are generated from the detection of at least one sensor (2), in particular a radar sensor, of the vehicle (1), wherein the object (3) is represented in the sensor data, Based on the provided sensor data, in the case of a machine learning model (50) being used, at least one representation of the state is ascertained (102), wherein the machine learning model (50) is trained for object detection and/or classification, -Based on the provided sensor data, in case at least one physical model is used, deriving (103) at least one further characterization of the state.
  2. 2. The method (100) according to claim 1, characterized in that the at least one representation is a class, a cartesian position, an orientation, a cartesian speed, a yaw rate and/or an extension of the object (3), and the state is preferably represented by a state vector.
  3. 3. The method (100) according to claim 2, characterized in that at least the class, cartesian position, orientation and extension of the object (3) is determined and provided in the form of a bounding box within the scope of the determination (102) using the machine learning model (50).
  4. 4. A method (100) according to claim 2 or 3, characterized in that at least the cartesian speed and/or the speed of the object (3) is/are determined within the scope of the determination (103) using the at least one physical model, preferably based on the bounding box provided.
  5. 5. The method (100) according to any one of the preceding claims, wherein the sensor data comprise a time stamp, which represents a point in time when the corresponding sensor data was detected, wherein the determination (102) using the machine learning model (50) and/or the determination (103) using the at least one physical model takes place taking into account the time stamp.
  6. 6. The method (100) according to any one of the preceding claims, wherein the method (100) further comprises: -performing object tracking of the object (3) using the determined state of the object (3), wherein the state of the object (3) is continuously updated using the machine learning model (50) and/or the at least one physical model within the scope of the object tracking.
  7. 7. The method (100) of claim 6, wherein the method (100) further comprises: Performing a maturity check based on the sensor data and/or on at least one representation determined using the machine learning model (50), wherein within the range of the maturity check the sensor data and/or at least one representation determined using the machine learning model (50) is checked with respect to at least one defined quality criterion, Wherein said object tracking is performed only if the result of said maturity inspection indicates that "said at least one defined quality criterion is met".
  8. 8. The method (100) according to any one of the preceding claims, wherein the method (100) further comprises: -initiating control of the vehicle (1) based on the determined state of the object (3).
  9. 9. A computer program (20) comprising instructions which, when the computer program (20) is implemented by a computer (10), cause the computer to implement the method (100) according to any of the preceding claims.
  10. 10. An apparatus (10) for data processing, arranged for implementing the method (100) according to any one of claims 1 to 8.
  11. 11. A computer-readable memory medium (15) comprising instructions which, when implemented by a computer (10), cause the computer to implement the steps of the method (100) according to any one of claims 1 to 8.

Description

Method, computer program, device and memory medium for ascertaining the state of an object in a vehicle environment Technical Field The invention relates to a method for determining the state of an object in the environment of a vehicle. Furthermore, the invention relates to a computer program, an apparatus and a memory medium for such a purpose. Background Auxiliary and (highly) automated driving functions utilize multiple sensor modalities to create images of the environment. This is called perception. Here, the environment includes a static world (infrastructure, vegetation, etc.) and moving objects (vehicles, pedestrians, etc.). In order to enable reliable sensing, sensors, such as radar sensors, are put into use. Advantageously, these sensors can also be used in particular at night, in fog, in water foam and in rain. They emit electromagnetic waves that reflect and are received again by the sensor. Thus, a localization (reflection, locations) is produced on the stationary object as well as on the moving object. The measurement space is in particular a polar space, and is located essentially with distance, radial relative speed and azimuth angle, and also elevation angle for the updated radar sensor. Additionally, an estimate is made of the radar cross section (simplified) that tells the reflectivity of the target. In order to be able to track objects in an environment stably and without interruption, a random observer, such as a kalman filter or a bernoulli filter, is generally used. In this regard, a distinction is made between two basic tasks. On the one hand, the initial conditions of the object have to be estimated (object creation/generation (Spawning)/tracking initialization (Track-initialization)), and on the other hand, object Maintenance (Track-Maintenance) has to be performed, i.e. the object has to be kept acquired over time and estimation errors have to be minimized here. In some cases, the required differentiation of these tasks is evident. In object creation for these tasks, the main emphasis is to avoid false positive tracking and false negative tracking, and to estimate the initial state as accurately as possible. Since the object may be tracked over a longer time frame, the emphasis on object maintenance is on correcting the current estimated state. Disclosure of Invention The subject matter of the present invention is a method according to the present invention, a computer program according to the present invention, an apparatus according to the present invention and a machine-readable storage medium according to the present invention. Further features and details of the invention are found in the description and the drawings. The features and details described in connection with the method according to the invention are of course also applicable here in connection with the computer program according to the invention, the device according to the invention and the computer-readable memory medium according to the invention, and vice versa, respectively, so that in the context of the invention all the time also mutual references are possible. The subject matter of the invention is in particular a method for determining the state of an object in the environment of a vehicle, comprising: -providing sensor data, wherein the sensor data is generated from the detection of at least one sensor of the vehicle, in particular a radar sensor, wherein the object is represented in the sensor data, wherein instead of a radar sensor, an imaging device sensor, an ultrasonic sensor or a lidar sensor is also conceivable; -based on the provided sensor data, determining at least one representation of the state using a machine learning model, in particular a (deep) neural network, wherein the machine learning model is trained for object detection and/or classification; -based on the provided sensor data, in case at least one physical model is used, deriving at least one further characterization of the state. The at least one physical model may be, for example, a kinematic model in which motion is determined based on speed, acceleration and/or direction changes. Also included are models that advantageously utilize, for example, rigid body characteristics of the object. Furthermore, the at least one physical model may be a kinetic model that also takes into account forces, friction or mass in order to provide a more complex motion description. The at least one physical model may also be a radar-based or lidar-based model in which signal propagation is used for distance estimation and velocity estimation. Furthermore, the at least one physical model may also be an optical flow model in which motion is determined from apparent object optical flow in the scene. It is also conceivable that the at least one physical model is a driving dynamics model which models the vehicle behavior of the vehicle itself, for example when cornering or when the roadway is wet. Thus, a more comprehensive and accurate state, in partic