Method for controlling autonomous agents using self-reinforcement

Patent metadata

Field	Value
Publication number	US-11435742-B2
Application number	US-201916717934-A
Country	US
Kind code	B2
Filing date	Dec 17, 2019
Priority date	Dec 18, 2018
Publication date	Sep 6, 2022
Grant date	Sep 6, 2022

Abstract

The behavior of automated agents, such as autonomous vehicles, drones, and the like, can be improved by control systems and methods that implement a combination of neighbor following behavior, or neighbor-averaged information transfer, with delayed self-reinforcement by utilizing time-delayed movement data to modify course corrections of each automated agent. Disclosed herein are systems and methods by which a follower agent, or a multiple follower agents in formation with a plurality of automated agents, can be controlled by generating course correction data for each follower agent based on the movement of neighboring agents in formation, and augmenting the course correction data based on time-delayed movement data of the follower agent. The delayed self-reinforcement behavior can (i) increase the information-transfer rate between autonomous agents without requiring an increased, individual update-rate; and (ii) cause superfluid-like information transfer between the autonomous agents, resulting in improvements in formation-keeping performance of the autonomous agents.

First claim

What is claimed is:

1. A system for controlling navigation of an automated vehicle in response to movement of adjacent automated vehicles, the system comprising:
an automated vehicle in formation with adjacent automated vehicles, the automated vehicle comprising a sensing device configured to detect movement of the adjacent automated vehicles; and
a controller comprising at least one processor and a memory device containing executable instructions that, when executed by the at least one processor, cause the controller to:
determine an average movement of a subset of the adjacent automated vehicles over a first period of time based on output of the sensing device;
generate course correction data based on the average movement of the subset of the adjacent automated vehicles over the first period of time;
determine time-delayed movement data corresponding to a prior change in position of the automated vehicle over a second period of time preceding the first period of time;
augment the course correction data based on the time-delayed movement data to generate augmented course correction data, wherein the augmented course correction data can be expressed as a function of a time derivative of a state z i (s) according to equation:

z
.

i

⁡

(
s
)

=

-
αβ

⁢

⁢

K
i

⁢

Z
⁡

(
s
)

+

αβ
⁢

⁢

B
iZs

⁡

(
s
)

+

(

1
-

e

-
τ

⁢

⁢
s

τ

)

⁢

v
i

⁡

(
s
)

.

wherein ż i (s) is the time derivative of the state z i (s), and v i (s) is a function of the time-delayed movement data; and
instruct the automated vehicle to move based on the augmented course correction data.

2. The system of claim 1 , wherein:
the adjacent automated vehicles comprise a convoy of autonomous ground vehicles configured to operate on a road; and
the automated vehicle is instructed to move so as to cause the automated vehicle to maintain formation with the convoy.

3. The system of claim 1 , wherein:
the adjacent automated vehicles comprises a formation of unmanned aerial vehicles configured to operate in a three-dimensional airspace; and
the automated vehicle is instructed to move so as to cause the automated vehicle to maintain formation with the formation.

4. A method for controlling navigation of an automated agent in response to movement of adjacent automated agents, the method comprising:
receiving movement data indicative of an average movement of a subset of the adjacent automated agents neighboring the automated agent, with respect to the automated agent over a first period of time;
generating course correction data based on the movement data;
determining time-delayed movement data corresponding to prior movement of the automated agent over a second period of time preceding the first period of time;
augmenting the course correction data based on the time-delayed movement data to generate augmented course correction data, wherein the augmented course correction data can be expressed as a function of a time derivative of a state z L (s) according to equation:

z
.

i

⁡

(
s
)

=

-
αβ

⁢

⁢

K
i

⁢

Z
⁡

(
s
)

+

αβ
⁢

⁢

B
iZs

⁡

(
s
)

+

(

1
-

e

-
τ

⁢

⁢
s

τ

)

⁢

v
i

⁡

Assignees

• Univ Washington

Inventors

• Devasia Santosh
• TIWARI anuj

Classifications

• G08G1/0129

  for creating historical data or processing based on historical data · CPC title

• G08G1/22 Primary

  Platooning, i.e. convoy of communicating vehicles · CPC title

• G08G5/57

  for unmanned aircraft · CPC title

• G08G5/55

  for a single aircraft · CPC title

• G08G5/26

  Transmission of traffic-related information between aircraft and ground stations · CPC title