Technologies for calibrating radars and tracking space objects. Some of such technologies enable a technique for calibrating a radar based on using -A- an elemental antenna (308), which can be embedded on a housing hosting a set of antenna elements, or -B- an antenna (146) mounted to a reflector. Some of such technologies enable a radar site containing a first 1D phased array (112) and a second 1D phased array (112), where the first 1D phased array sends a set of signals and receives a set of reflections based on the set of signals, and the second 1D phased array receives the set of reflections.

In one embodiment, a method of adjusting a speed limit of an ADV includes the operations of tracking objects within a field of view of the ADV; and identifying a set of stable objects from the objects tracked by the ADV based on a set of requirements. The method further includes the operations of identifying a subset of objects from the set of stable objects, the subset of objects having longest distances to the ADV; calculating a detection distance by averaging distances from the subset of stable obstacles to the ADV; and adjusting the speed limit of the ADV based on the detection distance using a predetermined algorithm.

An apparatus includes an inverse radar sensor model processor and a grid mapping processor. The inverse radar sensor model processor receives radar sensor data for a time k from a radar sensor, generates object data based on the radar sensor data, and calculates instantaneous masses at the time k for each cell in a field of view (FOV) of the radar sensor based on the object data and a sensor characteristic. The inverse radar sensor model processor outputs the calculated instantaneous masses to the grid mapping processor, which also receives accumulated masses for each cell in the FOV for a time period 0:k - 1. An accumulated mass represents a combination of instantaneous masses for the cell at each time increment in the time period 0:k - 1. The grid mapping processor generates updated accumulated masses for a time period 0:k.

Some radar sensors may provide a Doppler measurement indicating a relative velocity of an object to a velocity of the radar sensor. Techniques for determining a two-or-more-dimensional velocity from one or more radar measurements associated with an object may comprise determining a data structure that comprises a yaw assumption and a set of weights to tune the influence of the yaw assumption. Determining the two-or-more-dimensional velocity may further comprise using the data structure as part of regression algorithm to determine a velocity and/or yaw rate associated with the object.

A method of sensing a three-dimensional (3D) space using at least one sensor is proposed. The method can include acquiring spatial information over time for the sensed 3D space, applying a neural network based object classification model to the acquired spatial information over time to identify at least one object in the sensed 3D space. The method can also include tracking the sensed 3D space including the identified at least one object, and using information related to the tracked 3D space.

The present disclosure provides systems and techniques directed at object detection in a vehicle. In aspects, techniques include capturing current radar image data. The current radar image data includes at least one current point cloud. The current point cloud includes at least one current object point being related to an object, and each current object point includes spatial information related to the object. The techniques further include retrieving previous radar image data. The previous radar image data includes at least one previous point cloud. The previous point cloud includes at least one previous object point being related to the object, and each previous object point includes spatial information related to the object. The techniques further include concatenating the information from the current radar image data and the information from the previous radar image data to derive enhanced radar image data using a recurrent neural network.

A computer implemented method for object detection includes: determining a grid, the grid comprising a plurality of grid cells; determining, for a plurality of time steps, for each grid cell, a plurality of respective radar detection data, each radar detection data indicating a plurality of radar properties; determining, for each time step, a respective radar map indicating a pre-determined radar map property in each grid cell; converting the respective radar detection data of the plurality of grid cells for the plurality of time steps to a point representation of pre-determined first dimensions; converting the radar maps for the plurality of time steps to a map representation of pre-determined second dimensions, wherein the pre-determined first dimensions and the pre-determined second dimensions are at least partially identical; concatenating the point representation and the map representation to obtain concatenated data; and carrying out object detection based on the concatenated data.

Disclosed are aspects of methods performed by an onboard unit of a vehicle and aspects of apparatuses that include an onboard unit of a vehicle to perform the methods. An example of the method includes, for each of a plurality of vehicle positions on a road, determining a direct distance between the vehicle position and a position of a roadside unit. The method also includes determining, based on the determined direct distance, a planar distance between the vehicle position in a horizontal plane containing the vehicle position and a projection of a position of the roadside unit onto a plane of the road. The determining of the planar distance is performed based on a height of the roadside unit relative to a ground level of the road.

A method for operating radar sensors in a vehicle. At the outset, the acquired targets are divided into stationary targets and moving targets. The moving targets are then divided into primary targets, whereof the distances from the vehicle are less than a pre-definable threshold value, and secondary targets, whereof the distances from the vehicle are greater than a threshold value. The primary targets are fed to a first tracking device, which ascertains the states of the primary targets. The secondary targets are fed to a second tracking device, which ascertains the states of the secondary targets. The second tracking device carries out a computationally less powerful ascertainment of the states than the first tracking device.

A scan aggregator and filter for an autonomous vehicle includes a plurality of radar sensors, where each radar sensor performs a plurality of individual scans of a surrounding environment to obtain data in the form of a radar point cloud including a plurality of detection points. The scan aggregator and filter also includes an automated driving controller in electronic communication with the plurality of radar sensors. The automated driving controller is instructed to filter each of the individual scans to define a spatial region of interest and to remove the detection points of the radar point cloud that represent moving objects based on a first outlier-robust model estimation algorithm. The automated driving controller aggregates a predefined number of individual scans together based on a motion compensated aggregation technique to create an aggregated data scan and applies a plurality of density-based clustering algorithms to filter the aggregated data scan.