Granted Patent B1

US 12,187,287 · App. 17/702,534 · Granted Jan 7, 2025

Inventor: Taylor Andrew Arnicar (Sunnyvale, CA)

Assignee: Zoox, Inc.

B60W40/04B60W60/001G08G1/0112G08G1/0125G08G1/0137G08G1/052B60W2420/403B60W2420/408B60W2420/54B60W2554/4046B60W2556/45G08G1/0116G08G1/04

Abstract

Claims (85)

1. An autonomous vehicle system comprising:

a sensor array;

a memory cache;

one or more processors; and

non-transitory computer readable media storing instructions that when executed by the one or more processors, cause the autonomous vehicle system to perform operations comprising:

receiving, from the sensor array, first sensor data at a first time;

identifying, based at least on part on the first sensor data, a characteristic of an object detected in an environment;

generating, based at least in part on the characteristic, a behavior profile of the object;

determining a unique identifier corresponding to the object;

storing the behavior profile as being associated with the unique identifier in the memory cache;

causing, based at least in part on storing the behavior profile in the memory cache, a confidence score of the behavior profile to increase;

receiving, from the sensor array, second sensor data at a second time, the second sensor data including the characteristic of the object;

determining the unique identifier based at least in part on the characteristic;

accessing the behavior profile from the memory cache based on the unique identifier; and

altering a parameter of a control system of the autonomous vehicle system based at least in part on the behavior profile and the confidence score.

2. The autonomous vehicle system of claim 1 , further comprising a short-term memory cache, the operations further comprising:

determining a divergence score for the object representative of a likelihood that the object will perform an action that differs from a predicted action; and

storing the behavior profile in the short-term memory cache in response to determining that the divergence score meets or exceeds a divergence threshold.

3. The autonomous vehicle system of claim 1 , wherein the behavior profile is indicative of a likelihood for the object to take a particular action in response to a condition.

4. The autonomous vehicle system of claim 1 , wherein storing the behavior profile and the unique identifier is in response to the behavior profile comprising a divergence score and the divergence score exceeding a predetermined threshold.

5. The autonomous vehicle system of claim 1 , wherein the behavior profile and the unique identifier are associated with a geographic location, and wherein accessing the behavior profile is further based on the geographic location.

6. The autonomous vehicle system of claim 1 , wherein one or more objects identified by the sensor array have an initial behavior profile indicative of a neutral profile, the initial behavior profile adjusted based on observed behavior of the one or more objects.

7. A method, comprising:

receiving sensor data from one or more sensors communicably coupled to a vehicle, the sensor data comprising a characteristic associated with an object detected in an environment;

identifying a unique identifier associated with the object based at least in part on the characteristic;

determining a behavior profile associated with the object based at least in part on the behavior profile being associated with the unique identifier;

causing a confidence score of the behavior profile to increase; and

performing at least one of:

modifying a parameter of a control system of the vehicle based at least in part on the behavior profile and the confidence score; or

training a model using the behavior profile.

8. The method of claim 7 , further comprising

detecting a second object in the environment;

accessing a database of behavior profiles;

searching the database of behavior profiles based on a characteristic of the second object; and

based at least in part on a determination that a second behavior profile associated with the second object is not stored in the database, applying a default behavior profile for the second object.

9. The method of claim 7 , wherein the behavior profile comprises at least one of:

a first likelihood that the object will follow traffic laws;

a second likelihood of an abnormal action by the object; or

a third likelihood for the object to take an action based at least in part on a condition.

10. The method of claim 7 , further comprising:

storing the behavior profile in association with the unique identifier in a memory cache associated with the vehicle;

determining a period of time associated with storing behavior profiles in the memory cache, the period of time starting in response to the object no longer being detected;

determining that a current time is after the period of time associated with the behavior profile; and

removing the behavior profile and the unique identifier from the memory cache.

11. The method of claim 7 , further comprising:

storing the behavior profile in association with a geographic locator and the unique identifier of the object in a memory cache associated with the vehicle;

determining a location of the vehicle at a second time; and

removing the behavior profile and the unique identifier from the memory cache in response to a determination that a distance from the location to the geographic locator meets or exceeds a threshold distance.

12. The method of claim 7 , wherein identifying the unique identifier comprises generating a hash corresponding to the object, the hash generated based on one or more attributes of the object.

13. The method of claim 12 , further comprising:

determining a divergence score for the object representative of a likelihood that the object will perform an action that differs from a predicted action, wherein the action comprises at least one of:

a first traffic maneuver within a threshold distance of a second object;

a second traffic maneuver at a speed in excess of a threshold speed associated with the second traffic maneuver;

an abnormal traffic maneuver; or

a violation of a traffic law.

14. The method of claim 7 , further comprising:

determining a location of the vehicle; and

determining that the unique identifier is associated with the location of the vehicle, and wherein determining the behavior profile is further based on the location of the vehicle.

15. A non transitory computer readable medium storing instructions executable by a processor, wherein the instructions, when executed, cause the processor to perform operations comprising:

receiving sensor data from one or more sensors communicably coupled to a system of a vehicle, the sensor data comprising a characteristic associated with an object detected in an environment;

identifying a unique identifier associated with the object based at least in part on the characteristic;

determining a behavior profile associated with the object based at least in part on the behavior profile being associated with the unique identifier;

causing a confidence score of the behavior profile to increase; and

performing at least one of:

modifying a parameter of a control system of the vehicle based at least in part on the behavior profile and the confidence score; or

training a model using the behavior profile.

16. The non transitory computer readable medium of claim 15 , further comprising determining that a score associated with the behavior profile meets or exceeds a threshold score,

wherein modifying the parameter of the control system is based at least in part on the score meeting or exceeding the threshold score.

17. The non transitory computer readable medium of claim 15 , further comprising:

identifying an action associated with the object based at least in part on the sensor data;

determining that an action characteristic associated with the action is representative of abnormal behavior; and

in response to determining that the action characteristic is representative of abnormal behavior, generating the behavior profile associated with the object.

18. The non transitory computer readable medium of claim 17 , wherein the action comprises at least one of:

a first likelihood that the object will follow traffic laws;

a second likelihood of an abnormal action by the object; or

a third likelihood for the object to take an action in a traffic scenario.

19. The non transitory computer readable medium of claim 15 , further comprising:

storing the behavior profile in association with the unique identifier in a memory cache associated with the vehicle;

determining a period of time associated with storing behavior profiles in the memory cache, the period of time starting in response to the object no longer being detected;

determining that a current time is after the period of time associated with the behavior profile; and

removing the behavior profile and the unique identifier from the memory cache.

20. The non transitory computer readable medium of claim 15 , further comprising:

storing the behavior profile in association with a geographic locator and unique identifier of the object in a memory cache associated with the vehicle;

determining a location of the vehicle at a second time; and

removing the behavior profile and the unique identifier from the memory cache in response to a determination that a distance from the location to the geographic locator meets or exceeds a threshold distance.

Assignments (1)

References Cited (29)