An axis-misalignment estimation device estimates an axis-misalignment angle of a radar device mounted to a moving object. The axis-misalignment estimation device estimates the axis-misalignment angle using a plurality of different axis-misalignment angle estimation methods based on detection results of the radar device. The axis-misalignment estimation device determines whether a predefined employment condition is met, based on a plurality of axis-misalignment angle estimates estimated using respective ones of the plurality of axis-misalignment angle estimation methods. In response to determining that the employment condition is met, the axis-misalignment estimation device employs at least one of the plurality of axis-misalignment angle estimates.

In implementations of systems for estimating three-dimensional trajectories of physical objects, a computing device implements a three-dimensional trajectory system to receive radar data describing millimeter wavelength radio waves directed within a physical environment using beamforming and reflected from physical objects in the physical environment. The three-dimensional trajectory system generates a cloud of three-dimensional points based on the radar, each of the three-dimensional points corresponds to a reflected millimeter wavelength radio wave within a sliding temporal window. The three-dimensional points are grouped into at least one group based on Euclidean distances between the three-dimensional points within the cloud. The three-dimensional trajectory system generates an indication of a three-dimensional trajectory of a physical object corresponding to the at least one group using a Kalman filter to track a position and a velocity a centroid of the at least one group in three-dimensions.

The radio system (10) comprises a waveform generator (1) alternately generating an FMCW wave representing a linearly frequency-modulated continuous wave for radar imaging and a CW wave representing a wave kept at a given frequency for measuring a velocity vector, an amplification chain (2), a set (4) of transmit antennas (41, 42, 43), a set (5) of receive antennas (51, 52, 531, 532), a set (7) of receivers (71-2, 731, 732), and a signal processor (9) implementing processing operations on FMCW signals received from the one or more lateral antennas (51, 52) of the set (5) of receive antennas (51, 52, 531, 532) and spectrally analysing CW signals received from the one or more lateral antennas (51, 52) and from the one or more ventral antennas (531, 532) of the set (5) of receive antennas (51, 52, 531, 532) so as to supply SAR images and components of the velocity vector of said airborne vehicle (20).

In a case where a person moves away in a longitudinal direction from the vicinity of a vehicle, a condition for generating a warning is satisfied due to a change in a speed in a transverse direction. Thus, a warning system generates an erroneous warning. In order to solve this problem, there are proposed a target detection system for a vehicle and a target detection method for a vehicle, both of which are capable of computing a final risk level, taking into consideration not only results of recomputing a time-to-collision and an impact point, but also the presence or absence of a target that is detected by a camera sensor. The time-to-collision and the impact point are recomputed, taking into consideration a change in a speed in a transverse direction that occurs when the target moves in the longitudinal direction.

The techniques and systems herein enable target-velocity estimation using position variance. Specifically, a plurality of detections of a target are received for respective times as the target moves relative to a host vehicle. Based on the detections, two-dimensional positions of the target relative to the host vehicle are determined for the respective times. Based on the positions of the target at the respective times, a first variance is determined for a first dimension of the positions, and a second variance is determined for a second dimension of the positions. Based on the first and second variances, an estimated velocity of the target is calculated. By basing the estimated velocity on the variances of the positions, more-accurate estimated velocities may be generated sooner, thus enabling better performance of downstream operations.

A method for providing personal tracking data for controlling at least one function of a technical system. The method includes reading in, executing, and generating. During the reading in, sensor data from two spaced-apart radar sensors with partially overlapping fields of view are read in, the sensor data representing a point cloud made up of point targets that are detected with the aid of the radar sensors. During the execution, an estimation algorithm is executed, using the sensor data, in order to generate at least one corrected position profile track, the estimation algorithm applying a density-based cluster analysis algorithm. During the generating, the tracking data are generated using the at least one corrected position profile track.

An axial displacement angle estimation device repeatedly calculates an axial displacement angle based on the detection result of the radar apparatus. The axial displacement angle estimation device extracts the axial displacement angle included in a predetermined extraction angle range among a plurality of axial displacement angles, and calculate an average value and a median value of the extracted plurality of axial displacement angles to be an axial displacement angle average value and an axial displacement median value. The axial displacement angle estimation device determines, based on the axial displacement angle average value and the axial displacement angle median value, whether a predetermined allowable condition is met. The axial displacement angle estimation device utilizes, when determined that the predetermined allowable condition is met, the axial displacement angle average value as an estimation result of the axial displacement angle.

An example method to determine an object spin rate may include receiving a radar signal of a particular object in motion. The method may further include converting the radar signal into an input vector. The method may also include providing the input vector as input to a neural network. The neural network may include access to a set of initial data that has been generated based on multiple initial radar signals of multiple initial objects in motion. The method may further include determining a spin rate of the particular object in motion based on an analysis performed by the neural network of the input vector including time and frequency information of the particular object in motion in view of the set of initial data. The analysis may include comparing one or more elements of the input vector to one or more elements of the set of initial data.

Ground truth data may be too sparse to supervise training of a machine-learned (ML) model enough to achieve an ML model with sufficient accuracy/recall. For example, in some cases, ground truth data may only be available for every third, tenth, or hundredth frame of raw data. Training an ML model to detect a velocity of an object when ground truth data for training is sparse may comprise training the ML model to predict a future position of the object based at least in part on image, radar, and/or lidar data (e.g., for which no ground truth may be available). The ML model may be altered based at least in part on a difference between ground truth data associated with a future time and the future position.

A method for checking the plausibility of an initially known transverse movement of an object. The method includes: emission of a radar signal having constant signal frequency, and reception by a radar device of reflections of the radar signal having constant signal frequency; and checking the plausibility of the transverse movement of the object by analyzing frequency ranges corresponding to the transverse movement in a spectrum of the reflected radar signal having constant signal frequency.