Probabilistic modular lane transition state estimation

Patent metadata

Field	Value
Publication number	US-12056937-B2
Application number	US-202117525689-A
Country	US
Kind code	B2
Filing date	Nov 12, 2021
Priority date	Nov 12, 2021
Publication date	Aug 6, 2024
Grant date	Aug 6, 2024

Abstract

A system and method are provided for making a probabilistic determination of the state of a lane transition which may be controlled by a traffic signal, such as a traffic cue. A determination of the state of transition of a traffic signal may be used by a vehicle to determine a course of action to take. In making the determination, the states of multiple traffic signals may be combined into one collection of elements that has values associated with the likelihood of the traffic signals being in a given state. Optionally, a determination may be made of whether a traffic signal is occluded.

First claim

What is claimed is:

1. A system comprising:
a processor system having one or more processors; and
a memory system that stores one or more machine instructions, which when implemented cause the processor system to implement a method comprising:
detecting a plurality of traffic signals; and
determining a likelihood of a state of a traffic signal based on information characterizing a distribution of observations of the plurality of traffic signals, each of multiple traffic signals of the plurality of traffic signals being represented with a different collection of elements, and combining the different collections of elements into one collection of elements; and
determining an action for a vehicle to take based on the determining of the likelihood of the state.

2. The system claim 1 , the determining of the likelihood of the state of the traffic signal being a determination of a current likelihood of a current state, the determination of the current likelihood of the current state being further based on a prior likelihood of a prior state of the traffic signal.

3. The system claim 1 , the determining the likelihood of the state of a traffic signal comprising:
populating the elements of the different collection of elements with data based or observations of the multiple traffic signals;
determining a likelihood of a transition to a state based on the one collection of elements; and
determining an action for a vehicle to take based on the determining of the likelihood of the transition, where a given element of the different collection of elements is associated with a given traffic-signal state, and where values of individual elements of the one collection of elements are combinations of values of corresponding elements in the different collection of elements.

4. The system of claim 3 the different collections of elements comprising an element that is associated with whether a traffic signal of the plurality of traffic signals is occluded.

5. The system of claim 3 the different collections of elements comprising an element associated with whether a state of a traffic signal of the plurality of traffic signals is unknown.

6. The system of claim 1 , the method further comprising:
determining whether a particular traffic signal of the multiple traffic signals is occluded, and
determining how to treat information about the particular traffic signal when the particular traffic signal is occluded, based on the particular traffic light being occluded.

7. The system of claim 6 , the determining whether the particular traffic signal is occluded comprising:
building a three-dimensional model of a vicinity, the vicinity comprising a region of a road between the multiple traffic signals and the vehicle; and
determining whether an object interrupts a straight line connecting the vehicle to the particular traffic signal.

8. The system of claim 7 , the building of the three-dimensional model comprising constructing a bounding box for an object; and
the determining whether the object interrupts the straight-line comprising determining whether the straight line is interrupted by the bounding box.

9. The system of claim 6 , the determining of whether the particular traffic signal is occluded comprising:
retrieving information related to a location of the particular traffic signal from a map; and
determining whether the particular traffic signal is detected at the location.

10. The system of claim 9 , the determining of whether the particular traffic signal is found at the location comprising:
segmenting an image; and
determining whether a segment of the image that is associated with an object, other than the particular traffic signal, is in the location.

11. The system of claim 6 , the determining of how to treat information about the particular traffic signal that is occluded, comprising:
determining whether the particular traffic signal is fully occluded, and
if the particular traffic signal is fully occluded, ignoring sensor information related to a state of the particular traffic signal.

12. The system of claim 6 , the determining of how to treat information about the particular traffic signal that is occluded, comprising:
determining a likelihood that the particular traffic signal is occluded, and
adjusting a value associated with the particular traffic signal, so that an observation associated with the value is given less weight than were the particular traffic signal to have less of a likelihood of being occluded than the likelihood that was determined.

13. The system of the claim 1 , the determining of the likelihood of the state comprising:
determining a transition matrix comprising values representing a probability of a transition between two states of the traffic signal;
determining a likelihood of a lane transition state based on information about a prior likelihood of the state of the plurality of traffic signals and the transition matrix.

14. The system of the claim 1 , the distribution of observations being represented by an observation distribution matrix, the determining of the likelihood of the state comprising:
determining the observation distribution matrix, the observation distribution matrix comprising values associated with a current distribution of observations;
the determining of the likelihood of the state being further based on information about a likelihood of a prior state of the plurality of traffic signals.

15. The system of the claim 14 , elements of the observation distribution matrix comprising a product of
a likelihood that a given color recorded by a sensor corresponds to a particular color associated with a state of the traffic signal based on a measure of a similarity of
the given color recorded and
the particular color associated with the state of the traffic signal, and
a likelihood that the color received by the sensor corresponds to a particular state of the traffic signal based on a likelihood of color distortions.

16. The system of claim 15 , the color distortions being associated with environmental conditions.

17. The system of the claim 1 , the determining of the action for a vehicle to take comprising determining whether to wait for the plurality of traffic signals to have a state that grants permission to enter the intersection.

18. A system comprising:
a processor system having one or more processors; and
a memory system that stores one or more machine instructions, which when implemented cause the processor system to implement a method comprising:
determining a plurality of traffic signals at an intersection;
determining whether a traffic signal of the plurality of traffic signals is occluded;
determining elements of a vector that represent a likelihood of an observation of a state of the plurality of traffic signals, based on sensor information associated with the plurality of traffic signals and based on the determining of whether the traffic signal of the plurality of traffic signals is occluded;
determining a likelihood of a transition to a state of the plurality of traffic signals based on the vector; and
determining an action for a vehicle to take based on the determining of the likelihood of the transition.

19. The system of claim 18 , the determining whether a traffic signal of the plurality of traffic signals is occluded comprising:
building a three-dimensional model of a vicinity, the vicinity comprising a region of a road between the multiple traffic signals and the vehicle; and
determining whether any objects interrupt a straight line co

Assignees

• Toyota Res Inst Inc

Inventors

• Gong Peiyan
• Delp Michael J
• Potthast Christian

Classifications

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06T7/90

  Determination of colour characteristics · CPC title

• G06T2207/10024

  Color image · CPC title

• G06T2207/30252

  Vehicle exterior; Vicinity of vehicle · CPC title

• B60W40/04

  Traffic conditions · CPC title