CN-117222915-B - System and method for tracking an expanded state of a moving object using a composite measurement model

CN117222915BCN 117222915 BCN117222915 BCN 117222915BCN-117222915-B

Abstract

A tracking system for tracking an expanded state of an object is provided. The tracking system includes at least one processor and a memory storing instructions that, when executed by the at least one processor, cause the tracking system to perform a probability filter that iteratively tracks a confidence in an expanded state of an object, wherein the confidence is predicted using a motion model of the object and is further updated using a composite measurement model of the object. The composite measurement model includes a plurality of probability distributions constrained to lie on the contours of the object and has a predetermined relative geometric mapping to the center of the object. Further, the tracking system tracks the expanded state of the object based on the confidence of the update of the expanded state.

Inventors

WANG PU
YAO GANG
H. Munawar
K baintop
P Bufunuosi
Philip Olik

Assignees

三菱电机株式会社

Dates

Publication Date: 20260512
Application Date: 20220114
Priority Date: 20210505

Claims (15)

1. A tracking system (200) for tracking an expanded state of an object, the expanded state including a kinematic state indicative of one or a combination of a position and a velocity of a center of the object and an extended state indicative of one or a combination of a size and an orientation of the object, the tracking system comprising: At least one processor (204), and A memory (206), the memory (206) storing instructions that, when executed by the at least one processor (204), cause the tracking system (200) to: Receiving measurements associated with at least one sensor (202), wherein the at least one sensor (202) is configured to detect a scene comprising the object via one or more signal transmissions configured to generate one or more measurements of the object per transmission; Performing a probability filter that iteratively tracks a confidence level of the expanded state of the object, wherein the confidence level is predicted using a motion model (210 a) of the object and updated using a composite measurement model (210 b) of the object, wherein the composite measurement model (210 b) comprises a plurality of probability distributions constrained to lie on a contour of the object and has a predetermined relative geometric mapping to the center of the object, wherein in each iteration of iterative tracking the confidence level of the expanded state is updated based on a difference between a predicted confidence level and an updated confidence level, wherein the updated confidence level is estimated based on a probability of fitting the measurements of each of the plurality of probability distributions and mapped to the expanded state of the object based on a corresponding geometric mapping, and The expanded state of the object is tracked based on the updated confidence of the expanded state.
2. The tracking system of claim 1, wherein the processor is further configured to transform the composite measurement model from a unit coordinate system to a global coordinate system based on a confidence level of the prediction of the expanded state or an iteratively updated confidence level of an unscented transformation function.
3. The tracking system of claim 1, wherein the processor is further configured to assign the plurality of measurements to different ones of the plurality of probability distributions by treating the different ones as belonging to different objects.
4. The tracking system of claim 3, wherein the processor is further configured to assign the plurality of measurements to the different probability distributions as the different objects using probabilistic multi-hypothesis tracking (PMHT) to perform the assigning of the plurality of measurements to the different probability distributions.
5. The tracking system of claim 1, wherein the composite measurement model is learned based on a Expectation Maximization (EM) method.
6. The tracking system of claim 1, wherein one or more parameters of each probability distribution are represented by a Random Matrix Model (RMM) in a two-dimensional (2D) probability space.
7. The tracking system of claim 1, wherein the contour of the object corresponds to a B-spline curve.
8. The tracking system of claim 1, wherein a confidence of a prediction of a state of a vehicle is used to align the composite measurement model with the measurement using an unscented transformation.
9. The tracking system of claim 1, Wherein the processor is configured to determine a control input of a controller of a vehicle based on a tracked distended state of the object and control the vehicle in accordance with the control input, and Wherein the vehicle is operatively connected to the tracking system of claim 1.
10. A tracking method for tracking an expanded state of an object, wherein the expanded state comprises a kinematic state indicative of one or a combination of a position and a velocity of a center of the object and an extended state indicative of one or a combination of a size and an orientation of the object, the method comprising the steps of: Receiving measurements associated with at least one sensor, wherein the at least one sensor is configured to detect a scene comprising the object with one or more signal transmissions to generate one or more measurements of the object per transmission; Performing a probability filter that iteratively tracks a confidence level of the expanded state of the object, wherein the confidence level is predicted using a motion model of the object and updated using a composite measurement model of the object, wherein the composite measurement model comprises a plurality of probability distributions constrained to lie on a contour of the object and has a predetermined relative geometric mapping to the center of the object, wherein in each iteration of iterative tracking, the confidence level of the expanded state is updated based on a difference between a predicted confidence level and an updated confidence level, wherein the updated confidence level is estimated based on a probability of fitting the measurements of each of the plurality of probability distributions and mapped to the expanded state of the object based on a corresponding geometric mapping, and The expanded state of the object is tracked based on the updated confidence of the expanded state.
11. The tracking method of claim 10, further comprising transforming the composite measurement model from a unit coordinate system to a global coordinate system based on a confidence level of the prediction of the expanded state or an iteratively updated confidence level of an unscented transformation function.
12. The tracking method according to claim 10, wherein the tracking method further comprises the step of assigning the plurality of measurements to different ones of the plurality of probability distributions independently of each other by treating the different ones as belonging to different objects.
13. The tracking method of claim 12, further comprising the step of assigning the plurality of measurements to the different probability distributions as different objects using probabilistic multi-hypothesis tracking (PMHT) to assign the plurality of measurements to the different probability distributions.
14. The tracking method of claim 10, wherein the composite measurement model is learned based on a Expectation Maximization (EM) method.
15. A non-transitory computer readable storage medium having embodied thereon a program executable by a processor for performing a method for tracking an expanded state of an object, wherein the expanded state includes a kinematic state indicating one or a combination of a position and a velocity of a center of the object and an extended state indicating one or a combination of a size and an orientation of the object, the method comprising the steps of: Receiving measurements associated with at least one sensor, wherein the at least one sensor is configured to detect a scene comprising the object with one or more signal transmissions to generate one or more measurements of the object per transmission; Performing a probability filter that iteratively tracks a confidence level of the expanded state of the object, wherein the confidence level is predicted using a motion model of the object and updated using a composite measurement model of the object, wherein the composite measurement model comprises a plurality of probability distributions constrained to lie on a contour of the object and has a predetermined relative geometric mapping to the center of the object, wherein in each iteration of iterative tracking, the confidence level of the expanded state is updated based on a difference between a predicted confidence level and an updated confidence level, wherein the updated confidence level is estimated based on a probability of fitting the measurements of each of the plurality of probability distributions and mapped to the expanded state of the object based on a corresponding geometric mapping, and The expanded state of the object is tracked based on the updated confidence of the expanded state.

Description

System and method for tracking an expanded state of a moving object using a composite measurement model Technical Field The present disclosure relates generally to automotive object tracking, and more particularly, to a system and method for tracking an expanded state of an object using measurements of the object. Background Control systems employed by vehicles (e.g., autonomous and semi-autonomous vehicles) predict safe motion or paths for the vehicle to avoid collisions with obstacles such as other vehicles or pedestrians. In some scenarios, the vehicle is also configured to sense its surroundings, such as road edges, pedestrians, and other vehicles, by means of one or more sensors of the vehicle. Some of these sensors include ultrasonic sensors, cameras, and LIDAR sensors used in existing Advanced Driver Assistance Systems (ADAS). The control system of the vehicle tracks the object states (where the object states include kinematic states) of other vehicles based on the automotive radar measurements to control the vehicle. Extended Object Tracking (EOT) with multiple measurements per scan has shown improved object tracking by enhancing object states from only kinematic states to both kinematic states and extended states, as compared to conventional point object tracking which includes only one measurement per scan. The extended state provides the size and orientation of the tracked object. To achieve this, it is necessary to capture the spatial distribution (i.e. how the car radar measurements spatially distribute around the object) along with the sensor noise. The current approach includes a framework of fixed point sets on rigid bodies that require non-scalable data correlation between fixed point sets and automotive radar detection, even for single object tracking. Spatial models such as contour models and surface models bypass cumbersome data correlation steps. For automotive radar measurements, the profile model reflects a measured distribution along the profile of an object (e.g., a rigid body), and the surface model assumes that radar measurements are generated from the inner surface of a two-dimensional shape. Examples of contour models include simple rectangular shapes, and more general starburst shapes modeled by random hypersurface models or gaussian process models. Some surface models, such as gaussian-based ellipses and hierarchical gaussian-based ellipses models, are computationally much simpler than contour models that require more degrees of freedom to describe more complex shapes. However, the measurement of the object is affected by noise and only receives reflections from the object surface. Thus, the above model does not capture real world automotive radar measurements. Accordingly, there is a need for a system and method for tracking the kinematic and stretched state of an object by capturing real world automotive radar measurements. Disclosure of Invention It is an object of some embodiments to provide a system and method for tracking an expanded state of a subject. The expanded state of the object includes a kinematic state indicating one or a combination of position and velocity of the center of the object and an extended state indicating one or a combination of size and orientation of the object. The center of the object is one or a combination of an arbitrarily selected point, a geometric center of the object, a center of gravity of the object, a center of a rear axle of a wheel of the vehicle, and the like. A sensor (e.g., automotive radar) is used to track an object (e.g., a vehicle). In embodiments, an automotive radar may provide direct measurements of radial velocity, long operating range, small size of millimeter or sub-terahertz frequency bands, and high spatial resolution. In point object tracking, a single measurement per scan is received from a vehicle. Point object tracking provides only the kinematic state (position) of the vehicle. Furthermore, the vehicle is tracked using a probabilistic filter having a measurement model of the kinematic state distribution. In Extended Object Tracking (EOT), multiple measurements per scan are received. The plurality of measurements are constructed around the vehicle space. Extended object tracking provides both the kinematic state and the extended state of the vehicle. The vehicle is tracked using a probabilistic filter having a measurement model of the spread distribution. However, real world automotive radar measurement profiles indicate that multiple reflections from vehicles are complex. Because of this complexity, designing a correct measurement model becomes complex. Thus, conventional measurement models are only applicable to kinematic states, not to dilated states. To this end, in some embodiments, spatial models, such as contour models and surface models, are used to capture real world automotive radar measurements. However, the above spatial model is inaccurate. Some embodiments are based on the recognition that real world