A computer-implemented method for predicting properties of a plurality of objects in a vicinity of a vehicle includes multiple steps that can be carried out by computer hardware components. The method includes determining a grid map representation of road-users perception data, with the road-users perception data including tracked perception results and/or untracked sensor intermediate detections. The method also includes determining a grid map representation of static environment data based on data obtained from a perception system and/or a pre-determined map. The method further includes determining the properties of the plurality of objects based on the grid map representation of road-users perception data and the grid map representation of static environment data.

The described aspects and implementations enable fast and accurate verification of radar detection of objects in autonomous vehicle (AV) applications using combined processing of radar data and camera images. In one implementation, disclosed is a method and a system to perform the method that includes obtaining a radar data characterizing intensity of radar reflections from an environment of the AV, identifying, based on the radar data, a candidate object, obtaining a camera image depicting a region where the candidate object is located, and processing the radar data and the camera image using one or more machine-learning models to obtain a classification measure representing a likelihood that the candidate object is a real object.

In various examples, a deep neural network(s) (e.g., a convolutional neural network) may be trained to detect moving and stationary obstacles from RADAR data of a three dimensional (3D) space. In some embodiments, ground truth training data for the neural network(s) may be generated from LIDAR data. More specifically, a scene may be observed with RADAR and LIDAR sensors to collect RADAR data and LIDAR data for a particular time slice. The RADAR data may be used for input training data, and the LIDAR data associated with the same or closest time slice as the RADAR data may be annotated with ground truth labels identifying objects to be detected. The LIDAR labels may be propagated to the RADAR data, and LIDAR labels containing less than some threshold number of RADAR detections may be omitted. The (remaining) LIDAR labels may be used to generate ground truth data.

According to an example aspect of the present invention, there is provided a method comprising, transmitting by a wireless device, during a first phase, a first probe signal associated with a user and receiving a reflected version of the first probe signal, transmitting by the wireless device, during the first phase, the reflected version of the first probe signal to a ground truth classifier, transmitting by the wireless device, during a second phase, a second probe signal associated with the user and receiving a reflected version of the second probe signal and transmitting by the wireless device, during the second phase, the reflected version of the second probe signal to a trusted apparatus.

Deep learning to improve or gauge the performance of a surface-penetrating radar (SPR) system for localization or navigation. A vehicle may employ a terrain monitoring system including SPR for obtaining SPR signals as the vehicle travels along a route. An on-board computer including a processor and electronically stored instructions, executable by the processor, may analyze the acquired SPR images and computationally identify subsurface structures therein by using the acquired image as input to a predictor that has been computationally trained to identify subsurface structures in SPR images.

The present application provides a method, apparatus and electronic equipment for recognizing a posture of a target, a first receiving signal and a second receiving signal upon scattering of a transmitting signal from a target to be recognized are acquired, a first baseband signal is determined according to the first receiving signal and the transmitting signal, and a second baseband signal is determined according to the second receiving signal and the transmitting signal; and a category of the posture of the target to be recognized is finally determined according to the first baseband signal and the second baseband signal. The first baseband signal and the second baseband signal carry various feature values related to the posture of the target, including but not limited to transversal velocity information and radial velocity information, etc.

A processing load in a case where a plurality of different sensors is used can be reduced. An information processing apparatus according to an embodiment includes: a recognition processing unit (15, 40b) configured to perform recognition processing for recognizing a target object by adding, to an output of a first sensor (23), region information that is generated according to object likelihood detected in a process of object recognition processing based on an output of a second sensor (21) different from the first sensor.

In an embodiment, a method includes: receiving first radar data from a millimeter-wave radar sensor; receiving a set of hyperparameters with a radar processing chain; generating a first radar processing output using the radar processing chain based on the first radar data and the set of hyperparameters; updating the set of hyperparameters based on the first radar processing output using a hyperparameter selection neural network; receiving second radar data from the millimeter-wave radar sensor; and generating a second radar processing output using the radar processing chain based on the second radar data and the updated set of hyperparameters.

A method is described which includes receiving a point cloud having a plurality of data points each representing a 3D location in a 3D space, the point cloud being obtained using a detection and ranging (DAR) sensor. For each data point, associating the data point with a 3D volume containing the 3D location of the data point, the 3D volume being defined using a 3D lattice that partitions the 3D space based on spherical coordinates. For at least one 3D volume, the data points are sorted within the 3D volume based on at least one dimension of the 3D lattice; and the sorted data points are stored as a set of ordered data points. The method also includes performing feature extraction on the set of ordered data points to generate a set of ordered feature vectors and providing the set of ordered feature vectors to perform a machine learning inference task.

Systems and methods for training an artificial intelligence (AI) classifier of scanned items. The items may include a training set of sample raw scans. The set may include in-class objects and not-in-class raw scans. An AI classifier may be configured to sample raw scans in the training set, measure errors in the results, update classifier parameters based on the errors, and detect completion of training.