Get EVs & Auto every Monday

One free email with what the biggest companies in EVs & Auto filed, were granted and gave up this week.

The Thicket.
EVs & Auto / Week of Nov 16, 2026Get it every Monday →
All EVs & AutoIssue 001 / TeslaThe filing
TeslaGrant

Tesla secures grant for on-vehicle neural network trigger system to capture training data.

The filing

Tesla received a US grant on November 10, 2026 for a system that captures sensor data from a vehicle's surroundings to improve machine learning models. The company describes a problem where deep learning systems are limited by the quality and cost of manually curating training data, particularly for rare scenarios. The granted invention addresses this by running a first neural network on the vehicle to generate an intermediate result, then checking if that result meets a specific trigger condition indicating the network cannot make a predetermined classification. If the condition is met, a second neural network containing a trigger classifier, which was independently trained remotely, processes the intermediate result to determine a score for objects or scenarios like tunnel entrances or obstacles. When the score exceeds a threshold, the system transmits the sensor data and metadata to a network for use in retraining the neural network. The claims cover methods and systems where the trigger classifier is dynamically selected and transmitted wirelessly to the vehicle, with the broadest claim and further claims amended.

The claims

USPTO 12,792,972 / 3 independent claims of 20

  1. Claim 1Currently amended

    A method, comprising: receiving, by at least one processor, sensor data representing surroundings of a vehicle; executing, by the at least one processor, a first neural network to the sensor data to cause the first neural network to generate an intermediate result at an intermediate layer of the first neural network independent from outputs generated by trigger classifiers configured to receive the intermediate result as input, the first neural network configured to process a set of features of the surroundings of the vehicle to generate an final output that is not based on outputs generated by trigger classifiers; determining, by the at least one processor, that the intermediate result meets a first trigger condition of one or more trigger conditions indicating that the first neural network cannot make a predetermined classification; in response to determining that the intermediate result meets the first trigger condition, executing, by the at least one processor, a second neural network comprising a trigger classifier independently trained remote from the vehicle using previous intermediate results for a target use case to process the intermediate result as an input from the first neural network to cause the trigger classifier to determine a classifier score for the sensor data associated with the intermediate result, the classifier score indicating a real-world object, a scenario, or a feature in the surroundings of the vehicle; determining, by the at least one processor, to transmit via a computer network at least a portion of the sensor data associated with the intermediate result in response to determining the classifier score, wherein the determining to transmit includes comparing the classifier score with a threshold value; and transmitting, by the at least one processor, the at least a portion of the sensor data associated with the intermediate result for use as training data for retraining the first neural network or a different neural network. App. No. 17/250,825

  2. Claim 19Currently amended

    A computer program product, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for: receiving sensor data representing surroundings of a vehicle; executing a first neural network to the sensor data to cause the first neural network to generate an intermediate result at an intermediate layer of the first neural network independent from outputs generated by trigger classifiers configured to receive the intermediate result as input, the first neural network configured to process a set of features of the surroundings of the vehicle to generate an final output that is not based on outputs generated by trigger classifiers; App. No. 17/250,825 determining that the intermediate result meets a first trigger condition of one or more trigger conditions; in response to determining that the intermediate result meets the first trigger condition, executing, a second neural network comprising a trigger classifier independently trained remote from the vehicle using previous intermediate results for a target use case to process the intermediate result as an input from the first neural network to cause the trigger classifier to determine a classifier score the classifier score for the first sensor data , the classifier score indicating a real-world object, a scene associated with the intermediate result, or a feature in the surroundings of the vehicle; determining to transmit via a computer network at least a portion of the sensor data associated with the intermediate result in response to determining the classifier score, wherein determining to transmit includes comparing the classifier score with a threshold value; and transmitting the at least a portion of the sensor data associated with the intermediate result for use as training data for retraining the first neural network or a different neural network.

  3. Claim 20Currently amended

    A system, comprising: a sensor on a vehicle; an artificial intelligence processor; a vehicle control module; an image signal processor configured to: receive an image captured using the sensor representing surroundings of the vehicle; process the image to generate a processed image; and provide the processed image to a first neural network; a memory coupled with the artificial intelligence processor, wherein the memory is configured to provide the artificial intelligence processor with instructions which when executed cause the artificial intelligence processor to: receive the processed image; perform an inference using the first neural network on the processed image by executing the first neural network to using the processed image to cause the first neural network to generate an intermediate result at an intermediate layer of the first neural network and an App. No. 17/250,825 inference result that is not generated based on the intermediate result, the first neural network configured to process a set of features of the surroundings of the vehicle; determine that the intermediate result of the first neural network meets a first trigger condition of one or more trigger conditions; in response to determining that the intermediate result of the first neural network meets the first trigger condition, provide the intermediate result of the first neural network to a trigger classifier independently trained remote from the vehicle using previous intermediate results for a target use case to process the intermediate result as an input from the first neural network, wherein the trigger classifier is used to determine a classifier score corresponding to the image associated with the intermediate result, the classifier score indicating a real-world object, a scenario, or a feature in the surroundings of the vehicle; and provide an inference result of the first neural network to the vehicle control module to at least in part autonomously operate the vehicle; and a network interface configured to: transmit at least a portion of the image associated with the intermediate result based at least in part on the classifier score for retraining the first neural network or a different neural network.

The 17 dependent claims are in the record. Open the record (opens in a new tab)

Expert view

Tesla, Inc. received a US grant on November 10, 2026 for a system that captures sensor data from a vehicle's surroundings to improve machine learning models. The company describes a problem where deep learning systems are limited by the quality and cost of manually curating training data, particularly for rare scenarios. The granted invention addresses this by running a first neural network on the vehicle to generate an intermediate result, then checking if that result meets a specific trigger condition indicating the network cannot make a predetermined classification. If the condition is met, a second neural network containing a trigger classifier, which was independently trained remotely, processes the intermediate result to determine a score for objects or scenarios like tunnel entrances or obstacles. When the score exceeds a threshold, the system transmits the sensor data and metadata to a network for use in retraining the neural network. The claims cover methods and systems where the trigger classifier is dynamically selected and transmitted wirelessly to the vehicle, with the broadest claim and further claims amended.

Background

The applicant’s account of the prior art / USPTO 12,792,972

[4] Deep learning systems used for applications such as autonomous driving are developed by training a machine learning model. Typically, the performance of the deep learning system is limited at least in part by the quality of the training set used to train the model. In many instances, significant resources are invested in collecting, curating, and annotating the training data. The effort required to create the training set can be significant and is often tedious. Moreover, it is often difficult to collect data for particular use cases that a machine learning model needs improvement on.

Get EVs & Auto every Monday.

© 2026 Fearn AI, Inc. All rights reserved.

Information from the public patent record, not legal advice.

Big Law patents. In plain sight.

The all-in-one patent firm for startups. Former Big Law patent experts and PhDs, on AI that runs on our own GPUs.

Social

Newsletter

Stay updated on the future of intellectual property.

Fearn Legal, LLC is a patent prosecution firm, licensed to file and prosecute patents across the USPTO. The information provided here is for general informational purposes only and does not constitute legal advice. Viewing this information or contacting us through this website does not create an attorney-client or patent agent-client relationship. For personalized legal advice, please reach out to us directly to set up a legal consult or consult directly with a qualified attorney.

Big Law patents. In plain sight.

Social

Newsletter

Stay updated on the future of intellectual property.

Fearn Legal, LLC is a patent prosecution firm, licensed to file and prosecute patents across the USPTO. The information provided here is for general informational purposes only and does not constitute legal advice. Viewing this information or contacting us through this website does not create an attorney-client or patent agent-client relationship. For personalized legal advice, please reach out to us directly to set up a legal consult or consult directly with a qualified attorney.

Big Law patents. In plain sight.

The all-in-one patent firm for startups. Former Big Law patent experts and PhDs, on AI that runs on our own GPUs.

Social

Newsletter

Stay updated on the future of intellectual property.

Fearn Legal, LLC is a patent prosecution firm, licensed to file and prosecute patents across the USPTO. The information provided here is for general informational purposes only and does not constitute legal advice. Viewing this information or contacting us through this website does not create an attorney-client or patent agent-client relationship. For personalized legal advice, please reach out to us directly to set up a legal consult or consult directly with a qualified attorney.