SYSTEM AND METHOD FOR OCCUPANCY ESTIMATION

• US 20110007944A1
• Filed: 02/26/2008
• Published: 01/13/2011
• Est. Priority Date: 09/19/2007
• Status: Active Grant

• Current Assignee: UTC Fire & Security Corporation (Raytheon Technologies Corporation d/b/a RTX)
• Original Assignee: UTC Fire & Security Corporation (Raytheon Technologies Corporation d/b/a RTX)
• Inventors: Burlatsky, Sergei F., Atrazhev, Vadim, Vasilyev, Oleg A., Erikhma, Nikolay S., Tomastik, Robert N.
• Granted Patent: US 8,428,918 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06V 20/53 Recognition of crowd images, e.g. recognition of crowd congestion

Abstract

A system generates occupancy estimates based on a Kinetic-Motion (KM)-based model that predicts the movements of occupants through a region divided into a plurality of segments. The system includes a controller for executing an algorithm representing the KM-based model. The KM-based model includes state equations that define each of the plurality of segments as containing congested portions and uncongested portions. The state equations define the movement of occupants based, in part, on the distinctions made between congested and uncongested portions of each segment.

Claims

1. A system for estimating occupancy in a region defined by a plurality of segments, the system comprising:

a controller for executing an algorithm that generates occupancy estimations associated with the predicted movement of occupants within each of the plurality of segments, wherein the algorithm is a Kinetic Motion (KM)-based model that predicts occupant movement based on modeled distinctions between congested and uncongested portions of each segment; and

an output operably connected to the controller to communicate the occupancy estimates generated by the algorithm.

2. The system of claim 1, wherein the algorithm defines the predicted movement of occupants based on an egress-mode of evacuation.

3. The system of claim 1, wherein the algorithm is initialized based on an initial distribution of occupants within the region.

4. The system of claim 3, wherein the initial distribution of occupants within the region is based on statistical, simulated, or stored data regarding occupant location within the region.

5. The system of claim 1, wherein the KM-based model generates occupancy estimates based on previous occupancy estimates and one or more state equations that define the congested portion as a queue having a length dependent on a number of occupants located in the congested portion, and the propagation of vacancies through the queue in response to an occupant existing the congested portion and an occupant entering the queue from a non-congested portion or entrance.

6. The system of claim 5, wherein the state equations predict a number of occupants located in the queue, a number of occupants located in the uncongested portion of the segment, and a number of occupants entering the segment, wherein the number of occupants entering the segments is based on whether the entrance to the segment is modeled as congested or uncongested, wherein the number of occupants entering a congested portion of the segment is dependent on the modeled propagation of vacancies through the queue.

7. The system of claim 1, further including:

an input operably connected to receive sensor data from on or more sensor devices, wherein the algorithm executed by the controller further includes an extended Kalman filter that generates occupancy estimates based on a combination of the sensor data and the predicted movement of occupants generated by the KM-based model.

8. A system for estimating occupancy in a region, the system comprising:

an input operably connected to receive sensor data from one or more sensor devices;

a kinetic motion (KM)-based model that predicts the movement of occupants within a region divided into a plurality of segments, wherein the KM-based model predicts occupant movement based on modeled distinctions between congested and uncongested portions of each segment based on modeling of each segment as including a congested portion and an uncongested portion;

an occupancy estimator operably connected to the input, wherein the occupancy estimator executes an algorithm to generate corrected occupancy estimates for the region based on the received sensor data and predictions of occupant movement generated by the KM-based model; and

an output operably connected to the occupancy estimator to communicate the occupancy estimate generated by the occupancy estimator.

9. The system of claim 8, wherein the KM-based model generates state predictions based on previous corrected estimates generated by the occupancy estimator and one or more state equations defining the predicted movement of occupants.

10. The system of claim 9, wherein the one or more state equations generate state predictions regarding a number of occupants in the congested portion of each segment, a number of occupants in the uncongested portion of each segment, and a number of occupants moving between adjacent segments.

11. The system of claim 10, wherein the state equations related to the number of occupants in the uncongested portions of each segment further define the number of occupants located in a plurality of elementary cells defined within each segment, wherein the state equations predict occupant movement to the adjacent elementary cell in a defined time-step.

12. The system of claim 10, wherein state equations related to the number of occupants in the congested portion of each segment further define the congested portion as a queue having a length that is related to the number of occupants in the queue, and the propagation of vacancies through the queue in response to an occupant existing the congested area.

13. The system of claim 10, wherein the state equations related to the number of occupants moving between adjacent segments are based, in part, on whether the entrance to a segment is modeled as congested or uncongested, wherein the number of occupants moving through an entrance to a congested portion of the segment is based, in part, on the modeled propagation of vacancies through the queue to the entrance.

14. The system of claim 10, wherein the algorithm executed by the occupancy estimator is an extended Kalman filter that corrects the state estimates generated by the KM-based model with the sensor data provided by the one or more sensor devices.

15. The system of claim 8, wherein the KM-based model in combination with received sensor data, including five video cameras located between adjacent zones and three located at exits, provides an occupancy estimate with an error per room that is a 60% improvement over occupancy estimates generated only with the received sensor data.

16. The system of claim 8, wherein the KM-based model in combination with sensor data, including five video cameras located between adjacent zones, three video cameras located at exits, and motion detection sensors located in each room, provides an occupancy estimate with an error per room that is a 76% improvement over occupancy estimates generated only with the sensor data received from the video cameras alone.

17. The system of claim 8, wherein the corrected occupancy estimate is generated in real-time, wherein updates are provided at one-second intervals.

18. A method for estimating occupancy in a region divided into a plurality of segments, the method comprising:

modeling each of the plurality of segments as containing a congested portion and uncongested portion;

calculating model-based predictions of occupancy within the plurality of segments based on state equations that model occupant movement in the congested portion of each segment, occupant movement in the uncongested portion of each segment, and occupant movement between segments based, in part, on whether an entrance to a particular segment is modeled as congested or uncongested; and

providing the model-based predictions of occupancy as an output.

19. The method of claim 18, further including:

defining initial conditions associated occupant location within the region based on statistical, simulated, or stored data regarding occupant location within the region

20. The method of claim 18, wherein calculating model-based predictions of occupancy within the plurality of segments based on state equations that model occupant movement in the congested portion of each segment includes:

modeling the congested portion as a queue having a length defined by the number of occupants located in the congested portion, wherein occupants are modeled as entering the queue based on the distance the queue extends into a segment.

21. The method of claim 20, wherein calculating model-based predictions of occupancy within the plurality of segments based on state equations that model occupant movement in the uncongested portion of each segment includes:

modeling the uncongested portion as a plurality of elementary cells in which occupants are modeled as traversing one elementary cell in each time step, wherein occupants are modeled as being added to the queue when occupants located in the elementary cell adjacent a tail of the queue are advanced in the next time step.

22. The method of claim 21, wherein calculating model-based predictions of occupancy within the plurality of segments based on state equations that model occupant movement between segments based, in part, on whether an entrance to a particular segment is modeled as congested or uncongested includes:

generating predictions associated with occupant movement between segments based on a width of an entrance to a segment, density of occupants in an elementary cell when the entrance is to the uncongested portion of the segment, and predicted propagation of vacancies in the queue when the entrance is to the congested portion of the segment.

23. The method of claim 18, further including:

acquiring sensor data from one or more sensor devices; and

correcting the model-based prediction of occupancy with the acquired sensor data to generate corrected occupancy estimates.

24. The method of claim 23, wherein correcting the model-based prediction of occupancy with the acquired sensor data to generate corrected occupancy estimates includes:

applying an extended Kalman filter that generates a mean estimate and a covariance associated with the number of occupants in the congested portion of each segment, the number occupants in the uncongested portion of each segment, and the number of occupants moving between adjacent segments.

25. The method of claim 24, wherein the extended Kalman filter further generates a mean estimate and a covariance associated with the number of occupants in each elementary cell of each segment.

26. A system for estimating occupancy in a region divided into a plurality of segments, the system comprising:

means for modeling each of the plurality of segments as containing a congested portion and uncongested portion;

means for calculating model-based predictions of occupancy within the plurality of segments based on state equations that model occupant movement in the congested portion of each segment, occupant movement in the uncongested portion of each segment, and occupant movement between segments based, in part, on whether an entrance to a particular segment is modeled as congested or uncongested; and

means for providing the model-based predictions of occupancy as an output.

27. The system of claim 26, further including:

means for defining initial conditions associated occupant location within the region based on statistical, simulated, or stored data regarding occupant location within the region

28. The system of claim 26, wherein the means for calculating model-based predictions of occupancy within the plurality of segments based on state equations includes:

means for modeling the congested portion as a queue having a length defined by the number of occupants located in the congested portion, wherein occupants are modeled as entering the queue based on the distance the queue extends into a segment.

means for modeling the uncongested portion as a plurality of elementary cells in which occupants are modeled as traversing one elementary cell in each time step, wherein occupants are modeled as being added to the queue when occupants located in the elementary cell adjacent a tail of the queue are advanced in the next time step; and

means for generating predictions associated with occupant movement between segments based on a width of an entrance to a segment, density of occupants in an elementary cell when the entrance is to the uncongested portion of the segment, and predicted propagation of vacancies in the queue when the entrance is to the congested portion of the segment.

29. The system of claim 26, further including:

means for acquiring sensor data from one or more sensor devices; and

means for correcting the model-based prediction of occupancy with the acquired sensor data to generate corrected occupancy estimates by applying an extended Kalman filter that generates a mean estimate and a covariance associated with the number of occupants in the congested portion of each segment, the number occupants in the uncongested portion of each segment, and the number of occupants moving between adjacent segments.

30. A computer readable storage medium encoded with a machine-readable computer program code for generating thereof occupancy estimates for a region, the computer readable storage medium including instructions for causing a controller to implement a method comprising:

modeling each of the plurality of segments as containing a congested portion and uncongested portion;

calculating model-based predictions of occupancy within the plurality of segments based on state equations that model occupant movement in the congested portion of each segment, occupant movement in the uncongested portion of each segment, and occupant movement between segments based, in part, on whether an entrance to a particular segment is modeled as congested or uncongested; and

communicating model-based predictions of occupancy.