A learning method and a learning device for performing customized route planning by supporting reinforcement learning by using human travel data as training data

摘要

PROBLEM TO BE SOLVED: To provide a learning method and apparatus for optimizing route planning for each passenger, a test method using the same, and a testing apparatus. In a learning device, a process S01-1, which uses an adjusted reward network to generate first adjusted rewards by referring to information on an actual situation vector and an actual motion included in a traveling locus, The common reward module refers to the information on the actual situation vector and the actual motion to generate the first common reward, and the process S01-2 is used as a prediction network to refer to the actual situation vector and calculate the actual expected values. A step S01-2 of performing a process of generating, a learning device learns a parameter of the adjustment reward network by performing a step S02 of generating an adjustment reward loss and backpropagation with a first loss layer. The step S03 is performed. [Selection diagram] Fig. 3