| CPC G07C 5/02 (2013.01) [G06Q 30/0284 (2013.01)] | 18 Claims |
|
1. A vehicle behavior evaluation device, which evaluates a behavior of an own vehicle, the vehicle behavior evaluation device comprising:
a setting unit setting multiple possible behaviors of an own vehicle when the own vehicle travels along a planned route;
a function device generating traveling situation related reward data by learning, wherein the generating of traveling situation related reward data includes:
simulating multiple combinations of situations of the own vehicle and situations of a different vehicle under multiple different environments;
assigning a first reward to a first situation of the own vehicle in which the own vehicle succeeds in avoiding a contact with the different vehicle; and
assigning a second reward to a second situation of the own vehicle in which the own vehicle contacts with the different vehicle, the second reward being set to be lower than the first reward; and
an evaluation unit calculating, using the function device, a reward to each of the multiple possible behaviors of the own vehicle with consideration of a traveling state of the different vehicle existing around the own vehicle and evaluating each of the multiple possible behaviors of the own vehicle based on the calculated reward,
wherein
the function device generates the traveling situation related reward data by:
dividing a traveling situation of the own vehicle and the different vehicle into multiple steps and evaluating the traveling situation of the own vehicle in each of the multiple steps;
setting a first cost, as a penalty, to the traveling situation of the own vehicle when the contact between the own vehicle and the different vehicle occurs;
setting a second cost, as a penalty, to the traveling situation of the own vehicle in one step that is previous to an occurrence of the contact between the own vehicle and the different vehicle, the second cost being set to be increased with a decrease of duration taken by the own vehicle until the occurrence of the contact between the own vehicle and the different vehicle; and
setting, as the reward, a reciprocal of a sum of the first cost and the second cost, wherein
the function device further sets a third cost to the traveling situation of the own vehicle in each of the multiple steps, which are divided from the traveling situation of the own vehicle and the different vehicle,
the third cost is set to be increased with a decrease of a distance between the own vehicle and the different vehicle, and
the reward included in the traveling situation related reward data is calculated with consideration of the third cost.
|