System and method for tracking detected objects

#The subject matter described herein relates, in general, to systems and methods for tracking detected objects.

#The background description provided is to present the context of the disclosure generally. Work of the inventor, to the extent it may be described in this background section, and aspects of the description that may not otherwise qualify as prior art at the time of filing, are neither expressly nor impliedly admitted as prior art against the present technology.

#Some current vehicles have systems capable of detecting objects external to the vehicle, such as other vehicles, pedestrians, buildings, trees, and the like. Some objects, such as vehicles and pedestrians, are dynamic and are capable of movement. Other objects, such as buildings and trees, are static and are not capable of movement. Regarding dynamic objects, the tracking of the movement of the dynamic objects is important for downstream tasks, such as motion planning systems that plan the motion of the vehicle based on the position and predicted movements of objects.

#The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate various systems, methods, and other embodiments of the disclosure. It will be appreciated that the illustrated element boundaries (e.g., boxes, groups of boxes, or other shapes) in the figures represent one embodiment of the boundaries. In some embodiments, one element may be designed as multiple elements, or multiple elements may be designed as one element. In some embodiments, an element shown as an internal component of another element may be implemented as an external component and vice versa. Furthermore, elements may not be drawn to scale.

#FIG. 1 illustrates a vehicle incorporating an object tracking system.

#FIG. 2 illustrates a more detailed view of the object tracking system.

#Described is an object tracking system that is capable of tracking multiple objects. The object tracking system generally includes a feature fusion module, a distance combination module, and a track initialization module. At a high level, the feature fusion module combines features from objects detected within a point cloud and objects detected within a corresponding image to generate fused features. These fused features can include a number of different features extracted from the point cloud and the image to capture the appearance and geometric information of an object.

#The distance combination module receives the fused features generated by the feature fusion module at the present time, as well as fused features generated by the feature fusion module at a prior time. Using the fused features at two different moments in time, including a Mahalanobis distance, the distance combination module can generate a learned distance metric that generally represents how similar objects are at the present time with objects of the prior time. This similarity is useful in determining which objects in the present time should be associated with objects in the prior time to be able to track objects properly. Moreover, objects that are detected in the present time may have been previously detected and should be associated with those previous detections. However, objects that are detected in the present time that were not detected before should be initialized as new tracks.

#As such, a matching module, which may use a greedy match algorithm, determines which objects in the present time have a match with objects detected in the prior time. Objects that are matched will receive the appropriate tracking number previously used. Unmatched objects are provided to the track initialization module that will apply a new tracking identifier to the unmatched objects. The track initialization module may determine a confidence score to filter out unmatched objects that are false positives. Unmatched objects that have an appropriate confidence score will be provided a new tracking identifier.