Multi-agent simulation system

1 . A multi-agent simulation system that performs a simulation of a target world in which a plurality of agents interacting with each other exist,

the multi-agent simulation system comprising:

a plurality of agent simulators configured to perform simulations of the plurality of agents, respectively; and

a center controller configured to communicate with the plurality of agent simulators, wherein

a subject agent is the agent allocated for each of the plurality of agent simulators,

wherein the center controller comprises one or more processors and a memory storing instructions that implement a movement message dispatcher that relays the communication of the movement messages between all of the plurality of the agent simulators, a plurality of message filters respectively corresponding to the plurality of agent simulators and retaining reference information REF 1 regarding processing time intervals of the plurality of agent simulators, a plurality of receive queues and a plurality of send queues, and a forwarder that forwards necessary ones among the result messages stored in each receive queue to necessary send queues, and the center controller retains reference information REF 2 including spatial configuration information of the simulation target world and functional information that indicates a correspondence relationship between a distance and a delivery frequency,

each agent simulator includes a send/receive controller and a 3D physics engine, the 3D physics engine including a surrounding agent status updater that estimates the current status of the surrounding agent existing around the subject agent based on the movement message, and a subject agent status updater that outputs the movement message including the updated status of the subject agent;

the center controller is further configured to deliver at least one delivery message indicating a status of at least one another agent different from the subject agent to each of the plurality of agent simulators,

the each agent simulator is configured to:

determine a status of the subject agent by performing a simulation of the subject agent based on the status of said another agent indicated by the delivery message; and

send a result message indicating the status of the subject agent acquired as a result of the simulation to the center controller,

a plurality of operation modes of the center controller include:

a first operation mode that delivers the result message received from each of the plurality of agent simulators as a new delivery message without filtering; and

a second operation mode that selectively delivers the result message received from each of the plurality of agent simulators as a new delivery message,

when a number of result messages that the center controller receives per unit time is equal to or less than a threshold, the center controller selects the first operation mode, and

when the number of result messages that the center controller receives per unit time exceeds the threshold, the center controller selects the second operation mode.

2 . The multi-agent simulation system according to claim 1 , wherein

the plurality of agent simulators include:

a first agent simulator that performs a simulation of a first agent at a first processing time interval; and

a second agent simulator that performs a simulation of a second agent different from the first agent at a second processing time interval longer than the first processing time interval, and

in the second operation mode, the center controller is further configured to set a number of a second delivery message delivered to the second agent simulator per unit time to be smaller than a number of a first delivery message delivered to the first agent simulator per unit time.

3 . The multi-agent simulation system according to claim 1 , wherein

a space of the target world is divided into a plurality of partitions,

the plurality of partitions include a first partition, a second partition, and a third partition,

the center controller implements:

a first partition receive queue configured to store the result message received from an agent simulator that performs a simulation of an agent belonging to the first partition;

a second partition receive queue configured to store the result message received from an agent simulator that performs a simulation of an agent belonging to the second partition; and

a third partition send queue configured to store the delivery message to be delivered to an agent simulator that performs a simulation of an agent belonging to the third partition,

in the target world, a distance between the first partition and the third partition is greater than a distance between the second partition and the third partition, and

in the second operation mode, the center controller is further configured to set a frequency of forwarding the result message from the first partition receive queue to the third partition send queue to be lower than a frequency of forwarding the result message from the second partition receive queue to the third partition send queue.

4 . The multi-agent simulation system according to claim 1 , wherein

the plurality of agent simulators include:

a first agent simulator configured to perform a simulation of a first agent;

a second agent simulator configured to perform a simulation of a second agent; and

a third agent simulator configured to perform a simulation of a third agent,

in the target world, a distance between the first agent and the third agent is greater than a distance between the second agent and the third agent,

in the second operation mode, the center controller is further configured to:

receive first result messages regarding the first agent from the first agent simulator and deliver at least a part of the first result messages as the delivery message to the third agent simulator;

receive second result messages regarding the second agent from the second agent simulator and deliver at least a part of the second result messages as the delivery message to the third agent simulator; and

set a frequency of delivering the first result message to the third agent simulator to be lower than a frequency of delivering the second result message to the third agent simulator.