Granted Patent B2

US 12,553,723 · App. 18/034,459 · Granted Feb 17, 2026

Inventors: Erwin Biebl (Munich, DE); Rodrigo Perez (Munich, DE); Ralph Helmar Rasshofer (Aichach, DE)

Assignee: Bayerische Motoren Werke Aktiengesellschaft

G01C21/28G06N3/045G06F18/24G06V20/56

Abstract

Claims (35)

1 . A system for detecting surroundings of a vehicle, the system comprising:

a sensor module for acquiring surroundings data; and

a plurality of computing devices that are spatially separate from one another and that are configured to process the acquired surroundings data by way of an artificial neural network, wherein the artificial neural network comprises multiple layers of artificial neurons, and each of the computing devices implements a subset of the layers;

wherein one of the computing devices is configured to output data from an output layer of the subset of layers assigned to the one of the computing devices to a downstream computing device from the plurality of computing devices;

wherein the output data is received by the downstream computing device;

wherein the output data is further processed by the subset of layers of the artificial neural network assigned to the downstream computing device; and

wherein the artificial neural network is trained such that, with a given data amount of the acquired surroundings data to be processed, a smallest possible amount of the output data is transmitted from the one of the computing devices to the downstream computing device.

2 . The system according to claim 1 , wherein:

at least one of the computing devices is configured to output output data from an output layer of a subset of layers assigned thereto to a downstream computing device from the plurality of computing devices, and

the downstream computing device is configured to receive the output data and to further process the output data by way of the subset of layers assigned thereto.

3 . The system according to claim 1 , wherein at least one of the computing devices is integrated in the sensor module.

4 . The system according to claim 1 , wherein at least one of the computing devices is arranged in or on the vehicle.

5 . The system according to claim 1 , wherein at least one of the computing devices is arranged at a distance from the vehicle.

6 . The system according to claim 1 , further comprising an interface for wireless data transmission of an intermediate result between one of the computing devices and another of the computing devices.

7 . The system according to claim 1 , wherein:

the system comprises a plurality of sensor modules for acquiring the surroundings data,

the computing devices jointly implement a plurality of artificial neural networks, and

each of the networks of the plurality of networks is assigned to a respective sensor module from the plurality of sensor modules and is configured to process the surroundings data acquired by way of the assigned sensor module.

8 . The system according to claim 7 , wherein:

each of the sensor modules is assigned a computing device close to a sensor from the plurality of computing devices, and

each of the assigned computing devices implements a subset of the layers of the network assigned to the respective sensor module.

9 . The system according to claim 8 , wherein at least one computing device separate from the sensor modules from the plurality of computing devices implements subsets of the layers of the plurality of networks.

10 . A computer product comprising a non-transitory computer readable medium having stored thereon program code which, when executed on a processor, causes the system according to claim 1 to carry out a method of detecting the surroundings of the vehicle by:

acquiring the surroundings data by way of the sensor module; and

processing the acquired surroundings data by way of the artificial neural network.

11 . A method for detecting surroundings of a vehicle, the method comprising:

acquiring surroundings data by way of a sensor module;

processing the acquired surroundings data by way of an artificial neural network, wherein:

the network is implemented by a plurality of computing devices that are spatially separate from one another,

the network comprises multiple layers of artificial neurons, and

each of the computing devices implements a subset of the layers;

outputting, by way of one of the computing devices, output data from an output layer of the subset of layers assigned to the one of the computing devices to a downstream computing device from the plurality of computing devices;

receiving the output data by way of the downstream computing device;

further processing the output data by way of the subset of layers of the network assigned to the downstream computing device; and

training the network such that, with a given data amount of the acquired surroundings data to be processed, a smallest possible amount of the output data is transmitted from the one of the computing devices to the downstream computing device.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (18)