US 12,007,499 B2 · Sensor calibration in advanced driver-assistance system verification

Abstract

Systems and methods are provided for validating a sensor system associated with a vehicle. The vehicle is equipped with a set of replica sensors selected and positioned to emulate an optical path of a master set of sensors that has already been validated. The second vehicle is driven for a distance less than that required for validation of the sensor system to acquire a first set of data associated with the set of replica sensors and a second set of data associated with the sensor system. A machine learning model is trained on the first and second sets of data to provide an image transfer function. The image transfer function is applied to a set of master validation data produced during validation of the master vehicle to produce a set of transformed validation data. The new sensor system is validated with the set of transformed validation data.

Claims19

Dependent claims

A system for validating a sensor system associated with a first vehicle comprising:

the sensor system associated with the first vehicle;
a replica sensor system mounted on the first vehicle and configured to emulate an optical path of a master sensor system associated with a second vehicle, the master sensor system having an associated set of validation data;
and
a validation data generator configured to generate a transfer function between a set of images taken at the sensor system associated with the first vehicle and a set of images taken at the replica sensor system and apply the transfer function to the set of validation data associated with the master sensor system to provide a set of validation data for the sensor system associated with the first vehicle, the transfer function comprising one of a mathematical expression and a set of internal parameters for an artificial neural network.

1 more element

A method for validating a sensor system associated with a vehicle comprising:

equipping the vehicle with a set of replica sensors selected and positioned to emulate an optical path of a master set of sensors, associated with a master vehicle, that has already been validated;
driving the vehicle for a distance less than that required for validation of the sensor system to acquire a first set of data associated with the set of replica sensors and a second set of data associated with the sensor system;
training an artificial neural network on the first set of data and the second set of data;
providing a set of master validation data produced during validation of the master vehicle to the artificial neural network to produce a set of transformed validation data;
and validating the new sensor system with the set of transformed validation data.

2 more elements

A method for validating a sensor system associated with a vehicle comprising:

equipping the vehicle with a set of replica sensors selected and positioned to emulate an optical path of a master set of sensors, associated with a master vehicle, that has already been validated;
driving the vehicle for a distance less than that required for validation of the sensor system to acquire a first set of data associated with the set of replica sensors and a second set of data associated with the sensor system;
training a convolutional neural network on the first set of data and the second set of data;
providing each of a set of images comprising a set of master validation data to the convolutional neural network to produce a set of transformed images;
and
validating the new sensor system with the set of transformed images.

3 more elements

Dependencies are inferred from the claim text.

Sensor calibration in advanced driver-assistance system verification

Life of this patent

Sensor calibration in advanced driver-assistance system verification

Life of this patent

Abstract

Claims19