A review of methods to model route choice behavior of bicyclists: inverse reinforcement learning in spatial context and recursive logit
Used for route choice modeling by the transportation research community, recursive logit is a form of inverse reinforcement learning, the field of learning an agent’s objective by observing it’s behavior. By solving a large-scale system of linear equations it allows estimation of an optimal (negative) reward function in a computationally efficient way that performs for large networks and a large number of observations. In this paper we review examples of IRL models applied to real world travel trajectories and look at some of the challenges with recursive logit for modeling bicycle route choice in the city center area of Amsterdam.
|Inverse reinforcement learning, GPS trajectory, Route choice modeling, Maximum entropy, Dynamic discrete choice, Markov decision process, Dynamic programming, Recursive logit, Bicycle route behavior|
|3rd ACM SIGSPATIAL International Workshop on GeoSpatial Simulation, GeoSim 2020|
Koch, T, & Dugundji, E.R. (2020). A review of methods to model route choice behavior of bicyclists: inverse reinforcement learning in spatial context and recursive logit. In ACM SIGSPATIAL. doi:10.1145/3423335.3428165