Method for the improved detection of objects by a driver assistance system
The disclosure relates to a method for operating a driver assistance system of a motor vehicle. The method includes detecting a first data set of sensor data measured by a sensor device of the driver assistance program. The first data set of sensor data includes missing class allocation information, wherein the class allocation information relates to the objects represented by the sensor data. The method also includes pre-training a classification algorithm of the driver assistance system while taking into consideration the first data set in order to improve the object differentiation of the classification algorithm. The method further includes generating a second data set of simulated sensor data which includes at least one respective piece of class allocation information according to a specific specification. The method also includes training the classification algorithm of the driver assistance system while taking into consideration the second data set in order to improve an allocation assignment of the classification algorithm for objects differentiated by the classification algorithm. The method further includes improving the detection of objects, which are represented by additional measured sensor data, by the driver assistance system.
1. A method for operating a driver assistance system of a vehicle, comprising:
detecting, by a sensor device of the driver assistance system, a first data set of sensor data, wherein the first data set of sensor data comprises missing class allocation information, the class allocation information relating to an object represented by the sensor data;
pre-training a classification algorithm of the driver assistance system while taking into consideration the first data set of sensor data in order to improve object differentiation of the classification algorithm, wherein the pre-training comprises an unmonitored learning process;
generating a second data set of simulated sensor data which comprises at least one respective piece of class allocation information according to a specified specification;
training the classification algorithm of the driver assistance system while taking into consideration the second data set of simulated sensor data in order to improve an allocation assignment of the classification algorithm for the object differentiated by the classification algorithm, wherein the training comprises a monitored learning process; and
improving detection of the object, which is represented by additional measured sensor data, by the driver assistance system.
2. The method of claim 1 , wherein the sensor data comprise image data, distance data, acoustic data or vehicle-internal sensor data.
3. The method of claim 1 , wherein the pre-training and detecting are performed repeatedly, or the training and generating are performed repeatedly.
4. The method of claim 1 , further comprising:
prior to the pre-training, providing the classification algorithm for the detection of the object or driving situations when operating the driver assistance system.
5. The method of claim 1 , wherein the pre-training is performed before the training.
6. The method of claim 1 , wherein the pre-training comprises an unmonitored learning process, which uses deep learning algorithms or representation learning.
7. The method of claim 1 , wherein the training comprises a monitored learning process, which uses gradient descent.
8. The method of claim 1 , wherein the first data set of sensor data is expanded before the pre-training with simulated sensor data.
9. The method of claim 1 , wherein the classification algorithm is based on a neural network.
10. The method of claim 9 , wherein the pre-training only relates to a first portion of the neural network and the training relates to a larger second portion of the neural network or the entire neural network.