In-Lab Testing/Learning Multimodal Van-Train-Van Policy 

  • System is running as a real-time service 

  • Real phones book multimodal rides in parallel to multimodal "virtual" rides (simulated phones booking)

  • Real phones shows drivers "next best action" (what to do)

  • System learns best policy in "a semi-realistic" environment

  • End-to-end system is validated as a real service (using drives, vans, users, trains stochastic models)

Such a testing method enables to add new MaaS features quickly, safely with highly predictable results for very large scale services 

©2017 by