Adjusting weight of intensity in a PHD filter based on sensor track ID

A method for tracking multiple objects with a probabilistic hypothesis density filter, the method comprising:

• receiving a plurality of reflected signals in a plurality of sensors from multiple objects in an environment around a vehicle, wherein the sensors are onboard the vehicle;
• detecting the received reflected signals in the sensors;
• computing a plurality of measurements based on the detected reflected signals for the multiple objects;
• generating a plurality of intensities, each intensity of the plurality of intensities including a weight, a state mean vector, and a state covariance matrix of statistics of a track of a respective one of the objects, wherein each intensity has one or more track IDs associated therewith, wherein each track ID associated with a respective intensity corresponds to a measurement that contributed to that respective intensity;
• pruning the plurality of intensities to produce a reduced set of intensities, wherein pruning includes in the reduced set of intensities, any intensity in the plurality of intensities having a weight above a respective threshold for that intensity and excludes from the reduced set of intensities, any intensity having a weight below the respective threshold for that intensity,
• wherein the respective threshold for an intensity is selected based on which one or more track IDs are associated with the respective intensity;
• wherein the plurality of intensities include one or more of a predicted intensity, a measurement-to-track intensity, and a new intensity.

A tracking system for a vehicle, comprising:

• one or more processing devices onboard the vehicle;
• a plurality of sensors onboard the vehicle and operatively coupled to the one or more processing devices;
• and
• one or more data storage devices onboard the vehicle and including instructions which, when executed by the one or more processing devices, cause the one or more processing devices to track multiple objects in an environment around the vehicle with a probabilistic hypothesis density filter, wherein the instructions cause the one or more processing devices to:
• detect a first set of reflected signals received by the plurality of sensors from multiple objects in the environment around the vehicle;
• compute a first set of measurements based on the detected first set of reflected signals;
• generate a plurality of Tk track intensities, each Tk track intensity of the plurality of Tk track intensities including a weight, a state mean vector, and a state covariance matrix of statistics of a track of a respective object at time Tk, wherein each Tk track intensity has one or more first track IDs associated therewith, wherein each first track ID associated with a respective Tk track intensity corresponds to one of the first set of measurements that contributed to that respective Tk track intensity;
• generate a plurality of Tk+1 predicted intensities based on the Tk track intensities, the plurality of Tk+1 predicted intensities corresponding to time Tk+1, wherein each of the plurality of Tk+1 predicted intensities includes a predicted weight based on the weight of the Tk track intensity;
• detect a second set of reflected signals received by the plurality of sensors from the multiple objects in the environment around the vehicle;
• compute a second set of measurements based on the detected second set of reflected signals, wherein the plurality of sensors provide a plurality of second track IDs for the second set of measurements, wherein the second set of measurements correspond to time Tk+1;
• compare the plurality of second track IDs to each first track ID associated with each Tk track intensity;
• if all of the first track IDs associated with a given Tk track intensity match any of the plurality of second track IDs, multiply the predicted weight of the Tk+1 predicted intensity corresponding to the given Tk track intensity by a first value;
• if less than all of the first track IDs associated with a given Tk intensity match any of the plurality of second track IDs, multiply the predicted weight of the Tk+1 predicted intensity corresponding to the given Tk track intensity by a second value, wherein the second value is greater than the first value;
• prune the plurality of Tk+1 predicted intensities based on their respective predicted weight after multiplying with the first value or the second value;
• and
• generate a plurality of Tk+1 track intensities based on the Tk+1 predicted intensities that are not pruned, each Tk+1 track intensity of the plurality of Tk+1 track intensities corresponding to time Tk+1 and having an updated weight based on the predicted weight of the respective Tk+1 predicted intensity after multiplying.