Robert Boschgmbh

Kornel Istvan Kis

Peter Korosi-Szabo

Levente Kis

Attila Borcs

Geza Velkey

Laszlo Anka

Chun Yang

Adam Rigo

A control system and a method for predicting a location of dynamic objects, for example, of pedestrians, which are able to be detected by the sensors of a vehicle. The control system includes a multitude of sensors and a processing system, which is configured to combine with a first program the objects that are detected by the multitude of sensors to form an object list, each entry of the object list encompassing the location, a speed and an open route for each of the objects, and the object list including a time stamp; and to determine with a second program for at least a portion of the dynamic objects an additional object list from a predefined number of object lists, the additional object list including a time stamp for a future point in time and encompassing at least the location of the dynamic objects.

Location prediction for dynamic objects

A padding method for a convolutional neural network, in particular for concentrically configured data. The method includes receiving concentrically configured data of an object, the concentrically configured data correlating with an image, which was recorded concentrically to the object, deconvolving the concentrically configured data to form a data array, including real-coherent data on opposite sides of the data array, carrying out a convolution operation by using ring padding, in the case of ring padding, the real-coherent data of one side of the data array being utilized for padding the real-coherent data of a side of the data array opposite thereto, and/or vice versa.

Padding method for a convolutional neural network

A method for predicting whether another vehicle in the driving-environment of an ego-vehicle will execute a lane-change, based on observations of the driving-environment of the ego-vehicle, including: the observations are supplied to individual classificators; based on at least a portion of the observations, each individual classificator, in accordance with an individual instruction, ascertains an individual probability that the other vehicle will change lanes; the driving situation in which the ego-vehicle finds itself is classified as a whole by a situation classificator into one of several discrete classes; a record of weighting factors, assigned to the class into which the situation-classificator has classified the driving-situation, is ascertained, that indicates the relative weighting of the individual classificators for this driving situation; the individual probabilities are set off against the weighting-factors to form an overall probability that the other vehicle will change lanes. A method for training weighting-factors and related computer-program are described.

Predicting lane changes of other vehicles

A device and method for automatic image enhancement in vehicles, in particular land vehicles, including a camera to record a primary-image, and an image-processing-module to determine a resulting-image from the primary-image. The image-processing-module includes image-processing-filters, each configured to transform the primary-image in each case into an intermediate-image, an evaluation-module that outputs a quality-index for each of the intermediate-images transformed with the image-processing-filter, a selection-module that selects the intermediate-image having the highest quality index and outputs it as the resulting-image, and a learning-neural-network to learn, in a learning phase, for each primary-image the image-processing-filter, from the image-processing-filters, having the highest quality index of the intermediate-image, and after the learning phase, for each primary-image to select the image-processing-filter, from the image-processing-filters, having the highest quality index of the intermediate-image.

Device and method for automatic image enhancement in vehicles

A system for assessing and/or adapting an image recorded with a camera, based on a deep neural network. The deep neural network undergoes a training, the training of the deep neural network is carried out based on a loss function as a metric, the loss function is based on a structural similarity index, and the structural similarity index in the training is ascertained based on at least one input image of the underlying camera, an output image of the deep neural network, and in particular an adaptation method via the deep neural network and a target image as default.

Growing community of inventors

Tapolca, Hungary

Kornel Istvan Kis