Information processing system, information processing method, program, and vehicle for generating a first driver model and generating a second driver model using the first driver model

1. An information processing system comprising:

at least one processor;

at least one first detector that detects a first surrounding situation that belongs to a first vehicle type; and

at least one second detector that detects a second surrounding situation that belongs to a second vehicle type that is different from the first vehicle type,

wherein

the at least one first detector and the at least one second detector have different detection precisions, and

the at least one processor

generates at least one first travel history that includes the first surrounding situation detected by the at least one first detector, as a first travel situation of a first vehicle that belongs to the first vehicle type,

receives the at least one first travel history from each of n vehicles that belong to the first vehicle type, where n is an integer that is larger than or equal to two,

learns based on the at least one first travel history to build a first driver model that represents a relation between the first travel situation and a first behavior of each of the n vehicles that belong to the first vehicle type,

generates at least one second travel history that includes the second surrounding situation detected by the at least one second detector, as a second travel situation of a second vehicle that belongs to the second vehicle type,

receives the at least one second travel history from each of m vehicles that belong to the second vehicle type that is different from the first vehicle type, where m is an integer that is smaller than n,

performs transfer learning in which the at least one second travel history is used for the first driver model to build a second driver model that represents relation between the second travel situation and a second behavior of each of the m vehicles that belong to the second vehicle type, and

forecasts a first forecast behavior of the first vehicle that belongs to the first vehicle type based on the second driver model, when the second detector of the second vehicle type has the detection precision that is higher than that of the first detector of the first vehicle type.

2. The information processing system according to claim 1 , wherein

the at least one processor,

when a first format of a first parameter that represents the first surrounding situation detected by the at least one first detector is different from a second format of a second parameter that represents the second surrounding situation detected by the at least one second detector,

converts at least one of (i) the first format of the first parameter of the at least one first detector and (ii) the second format of the second parameter of the at least one second detector to generate the at least one first travel history and the at least one second travel history that each include the first travel situation and the second travel situation represented by a unified format.

3. The information processing system according to claim 2 , wherein

the at least one processor

converts the first format of the first parameter of the at least one first detector and the second format of the second parameter of the at least one second detector into a predetermined format.

4. The information processing system according to claim 1 , wherein

the at least one processor

forecasts a second forecast behavior of the second vehicle that belongs to the second vehicle type, and the second forecast behavior is the second behavior related to the second parameter that represents the second surrounding situation detected by the at least one second detector in the second driver model.

5. The information processing system according to claim 1 , wherein

the at least one processor includes:

first processors of first vehicles that belong to the first vehicle type;

second processors of second vehicles that belong to the second vehicle type; and

a third processor of a server,

wherein the first processors

transmit, to the server, the at least one first travel history generated by travel of the first vehicles that belong to the first vehicle type,

the second processors

transmit, to the server, the at least one second travel history generated by travel of the second vehicles that belong to the second vehicle type, and

the third processor

receives the at least one first travel history from the first vehicles that belong to the first vehicle type, and receives the at least one second travel history from the second vehicles that belong to the second vehicle type.

6. The information processing system according to claim 5 , wherein

the third processor

transmits the second driver model built to the second processors.

7. The information processing system according to of claim 1 , wherein

the at least one processor further

receives at least one third travel history from each of k vehicles that belong to a third vehicle type, where k is an integer that is larger than m,

learns based on the at least one third travel history to build a third driver model that represents a relation between a third travel situation and a third behavior of each of the k vehicles that belong to the third vehicle type, and

receives a selection of the third driver model,

wherein the second driver model is built by performing transfer learning in which the at least one second travel history is used for the third driver model.

8. The information processing system according to claim 7 , wherein

the at least one processor

determines whether a plurality of vehicle types that are predetermined and usable for transfer learning include the first vehicle type and the third vehicle type.

9. A vehicle comprising the information processing system according to claim 1 .

10. An information processing method using an information processing system, the information processing system including at least one first detector that detects a first surrounding situation that belongs to a first vehicle type with a first detection precision and at least one second detector that detects a second surrounding situation that belongs to a second vehicle type with a second detection precision, the first vehicle type is different from the second vehicle type, the at least one first detector and the at least one second detector have different detection precisions,

the information processing method comprising:

generating at least one first travel history that includes the first surrounding situation detected by the at least one first detector, as a first travel situation of a first vehicle that belongs to the first vehicle type;

receiving the at least one first travel history from each of n vehicles that belong to the first vehicle type, where n is an integer that is larger than or equal to two;

learning based on the at least one first travel history to build a first driver model that represents a relation between the first travel situation and a first behavior of each of the n vehicles that belong to the first vehicle type;

generating at least one second travel history that includes the second surrounding situation detected by the at least one second detector, as a second travel situation of a second vehicle that belongs to the second vehicle type;

receiving the at least one second travel history from each of m vehicles that belong to the second vehicle type that is different from the first vehicle type, where m is an integer that is smaller than n;

performing transfer learning in which the at least one second travel history is used for the first driver model to build a second driver model that represents relation between the second travel situation and a second behavior of each of the m vehicles that belong to the second vehicle type; and

forecasting a first forecast behavior of the first vehicle that belongs to the first vehicle type based on the second driver model, when the second detector of the second vehicle type has the detection precision that is higher than that of the first detector of the first vehicle type.

11. A non-transitory machine-readable recording medium that stores a program that causes a computer to execute:

causing at least one first detector to detect a first surrounding situation that belongs to a first vehicle type;

causing at least one second detector to detect a second surrounding situation that belongs to a second vehicle type that is different from the first vehicle type, wherein the at least one first detector and the at least one second detector have different detection precisions;

generating at least one first travel history that includes the first surrounding situation detected by the at least one first detector, as a first travel situation of a first vehicle that belongs to the first vehicle type;

receiving the at least one first travel history from each of n vehicles that belong to the first vehicle type, where n is an integer that is larger than or equal to two;

learning based on the at least one first travel history to build a first driver model that represents a relation between the first travel situation and a first behavior of each of the n vehicles that belong to the first vehicle type;

generating at least one second travel history that includes the second surrounding situation detected by the at least one second detector, as a second travel situation of a second vehicle that belongs to the second vehicle type;

receiving the at least one second travel history from each of m vehicles that belong to the second vehicle type that is different from the first vehicle type, where m is an integer that is smaller than n;

performing transfer learning in which the at least one second travel history is used for the first driver model to build a second driver model that represents relation between the second travel situation and a second behavior of each of the m vehicles that belong to the second vehicle type; and

forecasting a first forecast behavior of the first vehicle that belongs to the first vehicle type based on the second driver model, when the second detector of the second vehicle type has the detection precision that is higher than that of the first detector of the first vehicle type.