Dr. Jelena Fiosina


EC-Rider: Explainable  AI  methods  for  human-centric  ridesharing.

Since April 2019 I has been working as a Postdoctoral researcher in this project. My task is connected with investigation and application of methods of data analysis as well as explanation generation methods.

  • S. Kraus, A. Azaria, J. Fiosina, M. Greve, N. Hazon, L. Kolbe, T.Lembcke, J. P. Müller, S. Schleibaum, M. Vollrath. (2020) AI for Explaining Decisions in Multi-Agent Environments, Blue Sky Paper, AAAI 2020 : The Thirty-Fourth AAAI Conference on Artificial Intelligence 2020 (submited)

This project is supported by the Volkswagen Foundation. The consortium comprises five internationally renowned universitary working groups from Germany and Israel, covering the core scientific areas required for achieving the project goals.

With EC-RIDER, we propose such a human-centric AI-enabled approach towards sustainable ridesharing, e.g. when finding the best assignment of passengers to drivers/vehicles. By taking user goals, needs, preferences, constraints, decision patterns and behavior into account more accurately, we hope to create future shared mobility services and underlying dynamic management mechanisms (e.g. pricing and incentives), that are not only efficient, but also acceptable and attractive to city dwellers.

Such a human-centric approach to shared mobility services based on AI methods needs to address a number of unsolved research challenges: 


  • RC1. Understand and model human motivation and behavior (user satisfaction, trust, choice of mobility mode) in ridesharing applications while maintaining user privacy. 
  • RC2. Develop and evaluate novel business and operations models including pricing and incentive schemes, taking the models of RC1 into account, in order to achieve a satisfactory degree of social wel-fare. 
  • RC3. Develop and evaluate innovative algorithmic AI methods based on the models provided by RC1 and RC2 to find fair, socially sustainable, and efficient traffic assignments. 
  • RC4. Develop new methodologies for successfully creating and applying the AI methods of RC3. The critical success factor we will address is explainability of AI methods and algorithms to the human user.