Abstract: A method is provided for the efficient training of a neural network for an Automated Driving System (ADS) of a vehicle, utilizing a Central Processing Unit (CPU) to handle input/output (I/O) operations for reading and writing sensor data from a disk, and a Graphics Processing Unit (GPU) for performing the network training. The method includes monitoring a GPU utilization value, which reflects the current or queued workload of the GPU, to ensure continuous, high-throughput training. If the monitored GPU utilization value is above a threshold, then the method performs the steps of reading the sensor data from the disk using the CPU, transferring it to the GPU, and training the neural network accordingly. If value is below the threshold, then the method performs the steps of generating synthetic sensor data by using the GPU, and training the neural network based on the synthetic sensor data by the GPU.

Abstract: A method for generating prediction output for an Automated Driving System (ADS) of a vehicle is disclosed. The method includes obtaining, by one or more processors, a first sensor dataset from a first sensor and a second sensor dataset from a different sensor, each including information about a portion of a surrounding environment of the vehicle. The method further includes processing the first sensor dataset using a first ensemble of two or more encoder networks, each trained to output a first set of encoded features, and processing the second dataset using a second ensemble of two or more encoder networks, each trained to output a second set of encoded features. Then, one or more sets of the encoded features from the first and second datasets are fused using a fusion algorithm to output fused encoded features. A decoder network then generates a prediction output based on the fused encoded features.

Abstract: The present invention relates to a computer-implemented method and a computing device. The method includes obtaining a second dataset including a set of sensor data sequences with associated annotations generated by a first machine learning model trained to perform a perception. Each sensor data sequence includes sensor data samples depicting a physical environment over a plurality of time instances. Then training a second machine learning model, using the second dataset, to perform an augmented perception task. The method also includes fine-tuning, using a third dataset, the second machine learning model, to perform the perception task, wherein the third dataset includes sensor data samples depicting a physical environment and that are annotated for the perception task. The method also includes providing the fine-tuned second machine learning model as a model for annotating training data for subsequent training of a production model, of an automated driving system, to perform the perception task.

Abstract: A method for making perception predictions for a perception functionality in an automated driving system of a vehicle is disclosed. The method includes generating 2D position information of an image captured by a vehicle-mounted camera. The 2D position information indicates a position of each pixel out of a plurality of pixels of the image, or a position of each patch out of a plurality of patches of the image in the 2D reference frame of the image. Then, feeding the generated 2D position information, extrinsic parameters of the vehicle-mounted camera, intrinsic parameters of the vehicle-mounted camera, and distortion parameters of the vehicle-mounted camera to a multilayer perceptron which process the feed data and output 3D positional encodings. The method further includes feeding the image data and the 3D positional encodings to a transformer network for generating a prediction output in a 3D/2D reference frame of the vehicle.

Abstract: A method for determining an association of an object to a target object class, the object being present in a surrounding environment of a vehicle travelling on a road is disclosed. The method includes obtaining sensor data, the sensor data including one or more images of the surrounding environment of the vehicle. The method further includes determining a presence of the object in the surrounding environment of the vehicle based on the obtained sensor data and obtaining a finite sub-set of annotated images being representative of the target object class. The finite sub-set includes a number of annotated images smaller than a threshold value number. Further, the method includes determining a preliminary association between the at least one object and the target object class and when the preliminary association is determined, the method further includes producing an image label for the one or more images of the object.

Abstract: The present invention relates to a method for performing a perception task of an electronic device or a vehicle, using a plurality of neural networks trained to generate a perception output based on an input image, wherein at least two neural networks of the plurality of neural networks are different from each other. The method includes: for a time instance of a plurality of consecutive time instances: obtaining an image depicting a portion of a surrounding environment of the electronic device or the vehicle; processing the image associated with the time instance using a subset of neural network(s) to obtain a network output for the time instance; and determining an aggregated network output by combining the obtained network output for the time instance with network outputs obtained for a number of preceding time instances.