Granted Patent B2

US 12,646,415 · App. 18/477,897 · Granted Jun 2, 2026

Inventors: Behdad Chalaki (Ann Arbor, MI); Vaishnav Tadiparthi (Ann Arbor, MI); Hossein Nourkhiz Mahjoub (Ann Arbor, MI); Ehsan Moradi Pari (Ann Arbor, MI); Jovin Jeevan D'Sa (Ann Arbor, MI)

Assignee: Honda Motor Co., Ltd.

G08G1/22B60W30/165G05D1/0293G06N3/092B60W2554/4041B60W2554/4044B60W2554/4046B60W2554/80

Abstract

Claims (308)

1 . A computer-implemented method for swarm adaptation based on social value orientations, comprising:

calculating social value orientations for each one of a plurality of non-swarm vehicles on a roadway with a plurality of swarm vehicles of a swarm on the roadway that act collectively based on an instantaneous traffic scenario and a shared goal of the swarm, wherein each one of the plurality of non-swarm vehicles and the swarm vehicles of the swarm are roadway vehicles traveling in the same longitudinal direction on the roadway, and the plurality of non-swarm vehicles are not participating the swarm;

determining a total cost for each one of the plurality of non-swarm vehicles based on the social value orientations, wherein the total cost for each one of the plurality of non-swarm vehicles is based on an individual cost of the corresponding non-swarm vehicle and a joint cost of the corresponding non-swarm vehicle relative to the roadway vehicles other than the corresponding non-swarm vehicle, the individual cost is based on a state of the corresponding non-swarm vehicle and an action of the corresponding non-swarm vehicle, and the joint cost is based on a stacked state of the roadway vehicles other than the corresponding non-swarm vehicle and a stacked action of the roadway vehicles other than the corresponding non-swarm vehicle;

generating a global objective function based on a summation of the total cost of the plurality of non-swarm vehicles;

determining a swarm action for a swarm vehicle of the swarm by minimizing the total cost of the global objective function; and

causing the swarm vehicle to execute the swarm action.

2 . The computer-implemented method of claim 1 , wherein the social value orientations are calculated based on inverse reinforcement learning.

3 . The computer-implemented method of claim 1 , further comprising:

determining a social behavior for a swarm vehicle based on the global objective function.

4 . The computer-implemented method of claim 1 , wherein the social value orientation is a radial angle.

5 . The computer-implemented method of claim 1 , wherein the social value orientation calculated to result in a binary allocation of each one of the plurality of non-swarm vehicles as exhibiting an egotistic social value orientation or an altruistic social value orientation.

6 . The computer-implemented method of claim 1 , wherein the total cost for each one of the plurality of non-swarm vehicles is given by:

J

i

=

cos

⁢

φ

i

·

c

i

(

x

i

,

u

i

)

+

sin

⁢

φ

i

·

c

i

j

⁢

o

⁢

i

⁢

n

⁢

t

(

x

i

,

u

i

,

x

i

_

,

u

i

_

)

,

where J i is the total cost,

φ i is the social value orientation of the corresponding non-swarm vehicle,

c i (x i ,u i ) is the individual cost,

x i is the state of the corresponding non-swarm vehicle,

u i is the action of the corresponding non-swarm vehicle,

c

i

j

⁢

o

⁢

i

⁢

n

⁢

t

(

x

i

,

u

i

,

x

i

_

,

u

i

_

)

is the joint cost,

x

i

_

is the stacked state of the roadway vehicles other than the corresponding non-swarm vehicle, and

u

i

_

is the stacked action of the roadway vehicles other than the corresponding non-swarm vehicle.

7 . A non-transitory computer readable storage medium storing instructions that when executed by a computer having a processor to perform a method for swarm adaptation based on social value orientations, the method comprising:

calculating social value orientations for each one of a plurality of non-swarm vehicles on a roadway with a plurality of swarm vehicles of a swarm on the roadway that act collectively based on an instantaneous traffic scenario and a shared goal of the swarm, wherein each one of the plurality of non-swarm vehicles and the swarm vehicles of the swarm are roadway vehicles traveling in the same longitudinal direction on the roadway, and the plurality of non-swarm vehicles are not participating the swarm;

determining a total cost for each one of the plurality of non-swarm vehicles based on the social value orientations, wherein the total cost for each one of the plurality of non-swarm vehicles is based on an individual cost of the corresponding non-swarm vehicle and a joint cost of the corresponding non-swarm vehicle relative to the roadway vehicles other than the corresponding non-swarm vehicle, the individual cost is based on a state of the corresponding non-swarm vehicle and an action of the corresponding non-swarm vehicle, and the joint cost is based on a stacked state of the roadway vehicles other than the corresponding non-swarm vehicle and a stacked action of the roadway vehicles other than the corresponding non-swarm vehicle;

generating a global objective function based on a summation of the total cost of the plurality of non-swarm vehicles;

determining a swarm action for a swarm vehicle of the swarm by minimizing the total cost of the global objective function; and

causing the swarm vehicle to execute the swarm action.

8 . The non-transitory computer readable storage medium of claim 7 , wherein the social value orientations are calculated based on inverse reinforcement learning.

9 . The non-transitory computer readable storage medium of claim 7 , further comprising:

determining a social behavior for a swarm vehicle based on the global objective function.

10 . The non-transitory computer readable storage medium of claim 7 , wherein the social value orientation is a radial angle.

11 . The non-transitory computer readable storage medium of claim 7 , wherein the social value orientation calculated to result in a binary allocation of each one of the plurality of non-swarm vehicles as exhibiting an egotistic social value orientation or an altruistic social value orientation.

12 . The non-transitory computer readable storage medium of claim 7 , wherein the total cost for each one of the plurality of non-swarm vehicles is given by:

J

i

=

cos

⁢

φ

i

·

c

i

(

x

i

,

u

i

)

+

sin

⁢

φ

i

·

c

i

j

⁢

o

⁢

i

⁢

n

⁢

t

(

x

i

,

u

i

,

x

i

_

,

u

i

_

)

,

where J i is the total cost,

φ i is the social value orientation of the corresponding non-swarm vehicle,

c i (x i ,u i ) is the individual cost,

x i is the state of the corresponding non-swarm vehicle,

u i is the action of the corresponding non-swarm vehicle,

c

i

j

⁢

o

⁢

i

⁢

n

⁢

t

(

x

i

,

u

i

,

x

i

_

,

u

i

_

)

is the joint cost,

x

i

_

is the stacked state of the roadway vehicles other than the corresponding non-swarm vehicle, and

u

i

_

is the stacked action of the roadway vehicles other than the corresponding non-swarm vehicle.

13 . A system for swarm adaptation based on social value orientations, comprising:

a processor; and

a memory storing instructions that when executed by the processor cause the processor to:

calculate social value orientations for each one of a plurality of non-swarm vehicles on a roadway with a plurality of swarm vehicles of a swarm on the roadway that act collectively based on an instantaneous traffic scenario and a shared goal of the swarm, wherein each one of the plurality of non-swarm vehicles and the swarm vehicles of the swarm are roadway vehicles traveling in the same longitudinal direction on the roadway, and the plurality of non-swarm vehicles are not participating the swarm;

determine a total cost for each one of the plurality of non-swarm vehicles based on the social value orientations, wherein the total cost for each one of the plurality of non-swarm vehicles is based on an individual cost of the corresponding non-swarm vehicle and a joint cost of the corresponding non-swarm vehicle relative to the roadway vehicles other than the corresponding non-swarm vehicle, the individual cost is based on a state of the corresponding non-swarm vehicle and an action of the corresponding non-swarm vehicle, and the joint cost is based on a stacked state of the roadway vehicles other than the corresponding non-swarm vehicle and a stacked action of the roadway vehicles other than the corresponding non-swarm vehicle;

generate a global objective function based on a summation of the total cost of the plurality of non-swarm vehicles;

determine a swarm action for a swarm vehicle of the swarm by minimizing the total cost of the global objective function; and

cause the swarm vehicle to execute the swarm action.

14 . The system of claim 13 , wherein the social value orientations are calculated based on inverse reinforcement learning.

15 . The system of claim 13 , further comprising:

determining a social behavior for a swarm vehicle based on the global objective function.

16 . The system of claim 13 , wherein the social value orientation is a radial angle.

17 . The system of claim 13 , wherein the social value orientation calculated to result in a binary allocation of each one of the plurality of non-swarm vehicles as exhibiting an egotistic social value orientation or an altruistic social value orientation.

18 . The system of claim 13 , wherein the total cost for each one of the plurality of non-swarm vehicles is given by:

J

i

=

cos

⁢

φ

i

·

c

i

(

x

i

,

u

i

)

+

sin

⁢

φ

i

·

c

i

j

⁢

o

⁢

i

⁢

n

⁢

t

(

x

i

,

u

i

,

x

i

_

,

u

i

_

)

,

where J i is the total cost,

φ i is the social value orientation of the corresponding non-swarm vehicle,

c i (x i ,u i ) is the individual cost,

x i is the state of the corresponding non-swarm vehicle,

u i is the action of the corresponding non-swarm vehicle,

c

i

j

⁢

o

⁢

i

⁢

n

⁢

t

(

x

i

,

u

i

,

x

i

_

,

u

i

_

)

is the joint cost,

x

i

_

is the stacked state of the roadway vehicles other than the corresponding non-swarm vehicle, and

u

i

_

is the stacked action of the roadway vehicles other than the corresponding non-swarm vehicle.

Assignments (1)

Continuity (1)

References Cited (37)