Granted Patent B2

US 12,386,059 · App. 18/246,144 · Granted Aug 12, 2025

Inventors: Sebastian Muenzner (Ludwigsburg, DE); Alexandru Paul Condurache (Renningen, DE); Claudius Glaeser (Ditzingen, DE); Fabian Timm (Renningen, DE); Florian Drews (Renningen, DE); Florian Faion (Staufen, DE); Jasmin Ebert (Heidelberg, DE); Lars Rosenbaum (Lahntal, DE); Michael Ulrich (Stuttgart, DE); Rainer Stal (Zurich, CH); Thomas Gumpp (Dillingen, DE)

Assignee: ROBERT BOSCH GMBH

G01S13/867G01S13/865G06V20/58

Abstract

Claims (39)

1. A method for monitoring surroundings of a first sensor system, comprising the following steps:

providing a temporal sequence of data of the first sensor system for monitoring the surroundings;

generating an input tensor including the temporal sequence of data of the first sensor system, for a trained neural network, the neural network being configured and trained to identify, based on the input tensor, at least one subregion of the surroundings, to improve the monitoring of the surroundings using a second sensor system;

generating a control signal for the second sensor system, using an output signal of the trained neural network, to improve the monitoring of the surroundings in the at least one subregion using the second sensor system.

2. The method as recited in claim 1 , wherein the neural network is configured to process time series of data and/or to characterize a time-dependent state of the neural network.

3. The method as recited in claim 1 , wherein the input tensor of the neural network includes data of the second sensor system, from a time step preceding a current time step, in order to improve the monitoring of the surroundings of the first sensor system.

4. The method as recited in claim 1 , wherein the first sensor system is a passive sensor system, and the second sensor system is an active sensor system.

5. The method as recited in claim 1 , wherein the first sensor system is the same sensor system as the second sensor system, and in a subsequent time step, the control signal is used by the sensor system for the improved monitoring of the surroundings.

6. The method as recited in claim 1 , wherein the control signal forces the second sensor system to monitor the subregion of the surroundings, to improve the monitoring of the surroundings of the first sensor system.

7. The method as recited in claim 1 , wherein the neural network is trained by:

providing an input tensor for the neural network, including a second temporal sequence of data of the first sensor system for monitoring surroundings of the first sensor system, the input tensor of the neural network including data of the second sensor system from a time step preceding a current time step;

generating at least one object of the surroundings of the first sensor system using the neural network and the input tensor;

generating a control signal using the neural network and the input tensor;

comparing the generated at least one object to at least one correspondingly associated reference object;

generating data of the second sensor system based on the control signal, for a next time step; and

adapting the neural network to minimize, during the determination of the object of the surroundings, a deviation from the correspondingly associated reference object.

8. A method for training a neural network to generate a control signal for a second sensor system, comprising the following steps:

providing an input tensor for the neural network, including a temporal sequence of data of a first sensor system for monitoring surroundings of the first sensor system, the input tensor of the neural network including data of a second sensor system, from a time step preceding a current time step;

generating at least one object of the surroundings of the first sensor system using the neural network and the input tensor;

generating a control signal using the neural network and the input tensor;

comparing the generated at least one object to at least one correspondingly associated reference object;

generating data of the second sensor system based on the control signal, for a next time step; and

adapting the neural network to minimize, during the determination of the object of the surroundings, a deviation from the correspondingly associated reference object.

9. The method as recited in claim 8 , wherein the at least one object and the correspondingly associated reference object are, in each instance, an object list including at least one object of the surroundings of the first sensor system, and/or the at least one object and the correspondingly associated reference object are, in each instance, a high-resolution representation of the surroundings of the first sensor system.

10. The method as recited in claim 8 , wherein the at least one object of the object list is generated using an object detector.

11. The method as recited in claim 8 , wherein the temporal sequence of data of the first sensor system are data of an actual, first sensor system or simulated data for the first sensor system.

12. The method as recited in claim 8 , wherein the data of the second sensor system are generated using a high-resolution sensor system, and/or the data of the second sensor system are generated using a simulation program for simulating the second sensor system, and/or the data of the second sensor system are generated by the second sensor system in the surroundings of the first sensor system.

13. The method as recited in claim 8 , wherein the neural network is configured to process time series of data and/or to characterize a time-dependent state of the neural network, and/or the neural network is a recurrent neural network.

14. A neural network configured and trained to generate a control signal for a second sensor system, the neural network being trained by:

providing an input tensor for the neural network, including a temporal sequence of data of a first sensor system for monitoring surroundings of the first sensor system, the input tensor of the neural network including data of a second sensor system, from a time step preceding a current time step;

generating at least one object of the surroundings of the first sensor system using the neural network and the input tensor;

generating a control signal using the neural network and the input tensor;

comparing the generated at least one object to at least one correspondingly associated reference object;

generating data of the second sensor system based on the control signal, for a next time step; and

adapting the neural network to minimize, during the determination of the object of the surroundings, a deviation from the correspondingly associated reference object.

15. A monitoring device configured to monitor surroundings of a first sensor system, the monitoring device configured to:

provide a temporal sequence of data of the first sensor system for monitoring the surroundings;

generate an input tensor including the temporal sequence of data of the first sensor system, for a trained neural network, the neural network being configured and trained to identify, based on the input tensor, at least one subregion of the surroundings, to improve the monitoring of the surroundings using a second sensor system; and

generate a control signal for the second sensor system, using an output signal of the trained neural network, to improve the monitoring of the surroundings in the at least one subregion using the second sensor system.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (16)