Delphi Technologies, Inc.

Aptiv Technologies Limited

Aptiv Technologies Ag

Delco Electronics Corporation

Jan Konried Schiffmann

David Aaron Schwartz

Kumar Vishwajeet

Wenbing Dang

Nianxia Cao

Keerthi Raj Nagaraja

Franz P Schiffmann

Peter James Schubert

Edward J Wallner

Syed Asif Imran

Sandeep D Punater

Susan Chen

David John Nichols

Yu Liu

Susan Yu-Ping Chen

Dale Horan

Adrian Becker

Amith Somanath

Lawrence A Humm

Lisa R Hamilton

Sachal Bhagwan Gidwani

Simon Roesler

David William Braun

Lawrrence C Hau

Xumin Zhu

Henrik Clasen

This document describes techniques and systems related to tracking different sections of articulated vehicles. A vehicle uses a radar system that can discern between unarticulated vehicles and articulated vehicles, which by definition have multiple sections that can pivot in different directions for turning or closely following a curve. The radar system obtains detections indicative of another vehicle traveling nearby. When the detections indicate the other vehicle is articulated, the radar system tracks each identifiable section, rather than tracking all the sections together. A bounding box is generated for each identifiable section; the radar system separately and concurrently monitors a velocity of each bounding box. The multiple bounding boxes that are drawn enable the radar system to accurately track each connected section of the articulated vehicle, including to detect whether any movement occurs between two connected sections, for accurately localizing the vehicle when driving.

Tracking different sections of articulated vehicles

This document describes techniques and systems for a partially-learned model for speed estimates in radar tracking. A radar system is described that determines radial-velocity maps of potential detections in an environment of a vehicle. The model uses a data cube to determine predicted boxes for the potential detections. Using the predicted boxes, the radar system determines Doppler measurements associated with the potential detections that correspond to the predicted boxes. The Doppler measurements are used to determine speed estimates for the predicted boxes based on the corresponding potential detections. These speed estimates may be more accurate than a speed estimate derived from the data cube and the model. Driving decisions supported by the speed estimates may result in safer and more comfortable vehicle behavior.

Partially-learned model for speed estimates in radar tracking

This document describes methods and systems directed at history-based identification of incompatible tracks. The historical trajectory of tracks can be advantageous to accurately determine whether tracks originating from different sensors identify the same object or different objects. However, recording historical data of several tracks may consume vast amounts of memory or computing resources, and related computations may become complex. The methods and systems described herein enable a sensor fusion system of an automobile or other vehicle to consider historical data when associating and pairing tracks, without requiring large amounts of memory and without tying up other computing resources.

History-based identification of incompatible tracks

This document describes radar tracking with model estimates augmented by radar detections. An example tracker analyzes information derived using radar detections to enhance radar tracks having object measurements estimated from directly analyzing data cubes with a model (e.g., a machine-learning model). High-quality tracks with measurements to objects of importance can be quickly produced with the model. However, the model only estimates measurements for classes of objects its training or programming can recognize. To improve estimated measurements from the model, or even in some cases, to convey additional classes of objects, the tracker separately analyzes detections. Detections that consistently align to objects recognized by the model can update model-derived measurements conveyed initially in the tracks. Consistently observed detections that do not align to existing tracks may be used to establish new tracks for conveying more classes of objects than the model can recognize.

Radar tracking with model estimates augmented by radar detections

Kurtosis based pruning for a sensor-fusion system includes aligning candidate radar tracks with one of multiple candidate visual tracks. For each candidate vision track, a weight or other evidence of matching is assigned to each candidate radar track. An inverse of matching errors between each candidate vision track and each candidate radar track contributes to the evidence, which is normalized to produce, for each candidate vision track, a distribution associated with all candidate radar tracks. A Kurtosis or a shape of this distribution is calculated. Based on Kurtosis values, some candidate radar tracks are selected for matching and other remaining candidate radar tracks are pruned. The Kurtosis is used to determine how many candidates to retain and how many to prune.

Kurtosis based pruning for sensor-fusion systems

This document describes Kurtosis based pruning for sensor-fusion systems. Kurtosis based pruning minimizes a total quantity of comparisons performed when fusing together large sets of data. Multiple candidate radar tracks may possibly align with one of multiple candidate visual tracks. For each candidate vision track, a weight or other evidence of matching is assigned to each candidate radar track. An inverse of matching errors between each candidate vision and each candidate radar track contributes to this evidence, which may be normalized to produce, for each candidate vision track, a distribution associated with all candidate radar tracks. A Kurtosis or shape of this distribution is calculated. Based on the Kurtosis values, some candidate radar tracks are selected for matching and other remaining candidate radar tracks are pruned. The Kurtosis aids in determining how many candidates to retain and how many to prune. In this way, Kurtosis based pruning can prevent combinatorial explosions due to large-scale matching.

An illustrative example method of tracking an object includes detecting one or more points on the object over time to obtain a plurality of detections, determining a position of each of the detections, determining a relationship between the determined positions, and determining an estimated heading angle of the object based on the relationship.

Heading angle estimation for object tracking

An illustrative example method is for estimating a friction characteristic of a surface beneath a vehicle that has a plurality of wheels contacting the surface. The method includes determining a wheel speed of at least one of the wheels, determining a velocity of the at least one of the wheels separately from determining the wheel speed, determining a wheel slip of the at least one of the wheels based on the determined wheel speed and the determined velocity, and determining the friction characteristic based on the determined wheel slip. Determining the velocity separately from the wheel speed is accomplished using at least one detector that provides an output corresponding to a range rate, such as a RADAR or LIDAR detector.

Driving surface friction characteristic determination

An illustrative example method of classifying a detected object includes detecting an object, determining that an estimated velocity of the object is below a preselected threshold velocity requiring classification, determining a time during which the object has been detected, determining a first distance the object moves during the time determining a speed of the object from the first distance and the time, determining a second distance that a centroid of the detected object moves during the time, and classifying the detected object as a slow moving object or a stationary object based on a relationship between the first and second distances and a relationship between the estimated velocity and the speed.

Radar based tracking of slow moving objects

An illustrative example method of tracking a detected object comprises determining that a tracked object is near a host vehicle, determining an estimated velocity of the tracked object, and classifying the tracked object as frozen relative to stationary ground when the estimated velocity is below a preselected object threshold and a speed of the host vehicle is below a preselected host threshold.

Classifying potentially stationary objects tracked by radar

An illustrative example method of tracking a moving object includes determining an initial pointing angle of the object from a tracking device, determining an estimated position of a selected feature on the object based upon the initial pointing angle, determining a velocity vector at the estimated position, determining a lateral acceleration at the estimated position based upon the velocity vector and a yaw rate of the object, determining a sideslip angle of the selected feature based on the lateral acceleration, and determining a refined pointing angle of the object from the determined sideslip angle.

Method and system for determining the pointing angle of a moving object

An illustrative example method of tracking a moving object includes determining a heading angle of a centroid of the object from a tracking sensor, determining a raw difference value corresponding to a difference between a pointing angle of a selected feature on the object and the heading angle, wherein the raw difference is based on a trajectory curvature of the centroid from the tracking sensor and a distance between the centroid and the selected feature, determining a filtered difference between the pointing angle and the heading angle using a low pass filter, and determining the pointing angle by subtracting the filtered difference from the heading angle.

An object tracking system suitable for use on an automated vehicle includes a camera, a radar-sensor and a controller. The controller is configured to assign a vision-identification to each vision-track associated with an instance of an object detected using the camera, and assign a radar-identification to each radar-glob associated with an instance of grouped-tracklets indicated detected using the radar-sensor. The controller is further configured to determine probabilities that a vision-track and a radar-glob indicate the same object. If the combination has a reasonable chance of matching it is includes in a further screening of the data to determine a combination of pairings of each vision-track to a radar-track that has the greatest probability of being the correct combination.

Object tracking system with radar/vision fusion for automated vehicles

In accordance with one embodiment, a radar system with auto-alignment suitable for use in an automated vehicle is provided. The system includes a radar-sensor, a speed-sensor, and a controller. The radar-sensor is used to detect objects present in a field-of-view proximate to a host-vehicle on which the radar-sensor is mounted. The radar-sensor is operable to determine a measured-range-rate (dRm), a measured-azimuth-angle (Am), and a measured-elevation-angle (Em) to each of at least three objects present in the field-of-view. The speed-sensor is used to determine a measured-speed (Sm) of the host-vehicle. The controller is in communication with the radar-sensor and the speed-sensor. The controller is configured to simultaneously determine a speed-scaling-error (Bs) of the measured-speed, an azimuth-misalignment (Ba) of the radar-sensor, and an elevation-misalignment (Be) of the radar-sensor based on the measured-range-rate, the measured-azimuth-angle, and the measured-elevation-angle to each of the at least three objects, while the host-vehicle is moving.

Growing community of inventors

Newbury Park, CA, United States of America

Jan Konried Schiffmann