NON-KINEMATIC BEHAVIORAL MAPPING

• US 20100305858A1
• Filed: 06/01/2010
• Published: 12/02/2010
• Est. Priority Date: 06/01/2009
• Status: Active Grant

• Current Assignee: Raytheon Company (Rtx Corporation)
• Original Assignee: Raytheon Company (Rtx Corporation)
• Inventors: Richardson, James J.
• Granted Patent: US 8,538,675 B2
• Time in Patent Office: Days
• US Class Current: 701/301
• CPC Class Codes: G06N 7/02 using fuzzy logic computing arrangements based on biological models G06N3/00; computing arrangements using knowledge-based models G06N5/00
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G08G 1/166 for active traffic, e.g. moving vehicles, pedestrians, bikes

Abstract

A system and methodology/processes for non-kinematic/behavioral mapping to a local area abstraction (LAA) includes a technique for populating an LAA wherein human behavior or other non-strictly-kinematic motion may be present.

Claims

1. A system for predicting the path of a pedestrian comprising:

a local area abstraction (LAA) grid;

a pedestrian prediction logic (PPL) processor which receives a geographical map or image and operates on object data wherein human behavior, pedestrian behavior or other non-strictly-Kinematic motion may be present to populate the LAA grid.

2. The system of claim 1 wherein the PPL processor provides probabilistic object information to a pedestrian warning system (PWS).

3. The system of claim 2 further comprising a first pedestrian occupancy map (POM) which corresponds to a mini-map of a future positional probability distribution for an individual pedestrian or other non-strictly-Kinematic actor wherein given one or more parameters related to a pedestrian or an actor being modeled, including but not limited to positional location, speed, direction and size, said PPL computes probabilistic values and stores the values in said first POM.

4. The system of claim 3 wherein said first POM is a first one of a plurality of POMs, each of said plurality of POMs having stored therein values corresponding to a like plurality of individual pedestrians or other non-strictly-Kinematic actor.

5. The system of claim 4 wherein a single individual selected-POM is identified by performing a look-up or other procedure to determine the best/correct POM representation for a specific pedestrian or actor.

6. The system of claim 5 wherein for each POM chosen to model a current or future probable location(s) of a subject pedestrian or actor, the probability distribution of the individual POM is overlaid onto and added to an existing LAA.

7. The system of claim 6 wherein the LAA corresponds to an intersection occupancy map (IOM).

8. The system of claim 7 wherein the POM corresponds to one of:

a pin POM (PPOM);

or a cell POM (CPOM).

9. The system of claim 7 wherein the POM corresponds to a pin POM (PPOM) which assumes that the probabilities defined by the POM occur at some finite number of distinct points, each defined to be a specific distance and bearing from a reference position and a reference heading of the PPOM.

10. The system of claim 7 wherein the POM corresponds to a cell POM (CPOM) which assumes that the probabilities defined by the POM occur within some finite number of distinct cells.

11. The system of claim 7 wherein the POM is provided having a shape corresponding to one of:

a rectangular shape;

a square shape;

an oval shape;

a round shape;

a triangular shape;

a polygonal shape;

a convex polygonal shape;

a non-convex polygonal shape;

a non-polygonal shape;

a convex non-polygonal shape; and

a non-convex non-polygonal shape.

12. A method wherein populating a local area abstraction (LAA) comprises using probability values from a (pedestrian occupancy map (POM) to populate at least some cells of the LAA.

13. The method of claim 12 wherein the POM corresponds to a cell POM (CPOM) and the probability assigned to each cell/area of the CPOM is distributed proportionately to the LAA cell(s)/area(s) which the CPOM cells overlap.

14. The method of claim 13 wherein both the CPOM cells/areas and the LAA cells/areas are convex polygons and the method further comprises identifying as a convex polygon an area of overlap between the CPOM cells/areas and the LAA cells/areas.

15. The method of claim 14 wherein each of n points define a subject CPOM convex polygon and m points defining a subject LAA convex polygon and wherein the n point and m points are added to a list of points, L.

16. The method of claim 15 wherein for each of n sides of the CPOM convex polygon and m sides of the LAA convex polygon, an intersection of the CPOM and LAA is computed and wherein if the two sides intersect, the point of intersection is added to the list of points, L.

17. The method of claim 16 wherein for each point in the list of points L, checks are performed to determine if the point is on-or-within the CPOM cell area and/or on-or-within the LAA cell/area and wherein if the point is not on-or-within both the CPOM and LAA cells/areas, it is removed from the list of points, L and wherein at the conclusion of this operation, the list of points L contains only points that are on the perimeter of the area of overlap between the CPOM and LAA.

18. The method of claim 17 wherein if the number of points in list L is 0, 1, or 2, then the area of overlap=0.

19. The method of claim 18 wherein if the number of points in list L is greater than two, then processing continues as follows to place the points in list L into a clockwise order:

a. the first two points from list L are copied into clockwise list, C, in position numbers one and two;

b. for each of the p remaining points in list L, the point is inserted into successive trial slots in list C before, between and following the other points in list C and the resulting list is checked to see if it is counterclockwise at any point;

d. if the resulting list is found not to be counterclockwise at any point, then the test point has been properly inserted into list C and the next point in list L is processed;

e. If the resulting list is found to be counterclockwise at any point then the test point has not been properly inserted into list C and the next candidate slot is tried for the test point;

20. The method of claim 19 wherein a-e are repeated until all points in list L have been successfully inserted into list C and wherein the resulting final list C is a list of points defining the area of overlap between the convex CPOM and convex LAA polygons, sorted in clockwise order.

21. The method of claim 20 wherein the area of overlap defined by the clockwise list of points C and the area of the subject convex CPOM cell are calculated.

22. The method of claim 21 wherein the probability of the convex LAA cell/area is increased by the probability of the subject convex CPOM cell/area times the ratio of the overlap area defined by list C divided by the area of the subject convex CPOM polygon.

23. A method to perform computation-less collision detection, the method comprising:

generating an object occupancy map (OOM) having a plurality of cells;

generating an intersection occupancy map (IOM) having a plurality of cells;

generating a collision determination pattern vector (CDPV);

mapping probability values of cells in the OOM over the cells of the IOM;

for each IOM cell having a greater-than-zero probability value, assigning a value of one to a corresponding CDPV cell and assigning a value of zero to all other CPDV cells; and

identifying bit patterns contained in CDPV cells.

24. The method of claim 23 wherein the identified bit patterns are checked for collision alert thresholds.

25. The method of claim 23 further comprising:

populating the IOM cells with values corresponding to future projected object locations; and

setting a bit in the CDPV based upon each object'"'"'s classification type.

26. The method of claim 25 wherein a plurality of bits are used for each grouping and the CDPV reflects if zero, one or more objects of any type is predicted in an IOM cell at a future time.

27. The method of claim 23 wherein the OOM is provided as a pedestrian occupancy map (POM).

28. The method of claim 25 wherein identifying bit patterns contained in CDPV cells comprises performing a hashing operation of the CDPV to identify specific bit patterns.

29. An object warning system comprising:

a pedestrian warning system; and

a computation-less collision detection processor.

30. The object warning system of claim 29 wherein said computation-less collision detection processor comprises:

an object occupancy map (OOM) having a plurality of cells;

an intersection occupancy map (IOM) having a plurality of cells having stored therein probability values of cells from the OOM; and

a collision determination pattern vector (CDPV) having a plurality of cells wherein for each IOM cell having a greater-than-zero probability value, a corresponding CDPV cell has assigned thereto a first value and wherein all other CPDV cells have assigned thereto a second different value; and

an identification processor for identifying bit patterns contained in CDPV cells.