Arousal level control apparatus, arousal level control method, and recording medium
An arousal level control apparatus calculates a setting value of a control device under a constraint condition using an arousal level optimization model so that a value of an objective function is maximized. The control device affects a physical quantity of a surrounding environment that affects arousal level of a subject. The arousal level optimization model includes the constraint condition and the objective function. The constraint condition includes a physical quantity prediction model, an arousal level prediction model, and a setting value range condition that the setting value is within a predetermined range. The physical quantity prediction model is an explicit function that includes the physical quantity and the setting value as explanatory variables and has a predicted value of the physical quantity as an explained variable.
1. An arousal level control apparatus comprising:
at least one memory configured to store instructions; and
at least one processor configured to execute the instructions to:
calculate a setting value of a control device under a constraint condition using an arousal level optimization model so that a value of an objective function is maximized; and
set the control device with the calculated setting value, wherein
the control device is configured to affect a physical quantity of a surrounding environment that affects an arousal level of a subject,
the arousal level optimization model includes the constraint condition and the objective function,
the constraint condition includes a physical quantity prediction model, an arousal level prediction model, and a setting value range condition in which the setting value is within a predetermined range,
the physical quantity prediction model is an explicit function that includes the physical quantity and the setting value as explanatory variables and has a predicted value of the physical quantity as an explained variable,
the physical quantity prediction model is configured to be trained by machine learning,
the arousal level prediction model is an explicit function that includes the physical quantity and variation thereof over time as explanatory variables and has a predicted value of variation of the arousal level over time as an explained variable,
the arousal level prediction model is configured to be trained by machine learning,
the objective function expresses a total value or an average value of a predicted value for one or more subjects including the subject, the predicted value for the one or more subjects is a predicted value of variation in an arousal level for the one or more subjects and two or more time steps that satisfy a predetermined condition, and
the physical quantity includes a room temperature.
2. The arousal level control apparatus according to claim 1 , wherein the subject includes a vehicle driver.
3. The arousal level control apparatus according to claim 1 , wherein the constraint condition includes a comfort condition in which a sum of comfort penalty scores calculated for a plurality of setting values, including the setting value, is within a predetermined range.
4. The arousal level control apparatus according to claim 1 , wherein the arousal level prediction model includes variation of the arousal level over time as an explanatory variable.
5. The arousal level control apparatus according to claim 1 , wherein the arousal level prediction model includes temporal variance of the arousal level as an explanatory variable.
6. An arousal level control method executed by a computer, comprising:
calculating a setting value of a control device under a constraint condition using an arousal level optimization model so that a value of an objective function is maximized; and
setting the control device with the calculated setting value, wherein
the control device is configured to affect a physical quantity of a surrounding environment that affects arousal level of a subject,
the arousal level optimization model includes the constraint condition and the objective function,
the constraint condition includes a physical quantity prediction model, an arousal level prediction model, and a setting value range condition in which the setting value is within a predetermined range,
the physical quantity prediction model is an explicit function that includes the physical quantity and the setting value as explanatory variables and has a predicted value of the physical quantity as an explained variable,
the physical quantity prediction model is configured to be trained by machine learning,
the arousal level prediction model is an explicit function that includes the physical quantity and variation thereof over time as explanatory variables and has a predicted value of variation of the arousal level over time as an explained variable,
the arousal level prediction model is configured to be trained by machine learning,
the objective function expresses a total value or an average value of a predicted value for one or more subjects including the subject, the predicted value for the one or more subjects is a predicted value of variation in an arousal level for the one or more subjects and two or more time steps that satisfy a predetermined condition, and
the physical quantity includes a room temperature.
7. A non-transitory recording medium that stores a program causing a computer to execute:
calculating a setting value of a control device under a constraint condition using an arousal level optimization model so that a value of an objective function is maximized; and
setting the control device with the calculated setting value, wherein
the control device is configured to affect a physical quantity of a surrounding environment that affects arousal level of a subject,
the arousal level optimization model includes the constraint condition and the objective function,
the constraint condition includes a physical quantity prediction model, an arousal level prediction model, and a setting value range condition in which the setting value is within a predetermined range,
the physical quantity prediction model is an explicit function that includes the physical quantity and the setting value as explanatory variables and has a predicted value of the physical quantity as an explained variable,
the physical quantity prediction model is configured to be trained by machine learning,
the arousal level prediction model is an explicit function that includes the physical quantity and variation thereof over time as explanatory variables and has a predicted value of variation of the arousal level over time as an explained variable,
the arousal level prediction model is configured to be trained by machine learning,
the objective function expresses a total value or an average value of a predicted value for one or more subjects including the subject, the predicted value for the one or more subjects is a predicted value of variation in an arousal level for the one or more subjects and two or more time steps that satisfy a predetermined condition, and
the physical quantity includes a room temperature.