BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Collective and Strategic Routing of AV Fleets in Urban Networks
DTSTART:20261118T110000
DTEND:20261118T120000
DTSTAMP:20261109T145509Z
UID:990ad829ac0b514da80e18338c467950a1494a856dcc59c1002f87c3
CATEGORIES:Conferences - Seminars
DESCRIPTION:Rafał Kucharski (www.rafalkucharskilab.pl )  is an Associate
  Professor at the Faculty of Mathematics and Computer Science at Jagiellon
 ian University in Krakow\, Poland\, and a member of the Group of Machine L
 earning Research (GMUM). He leads the ERC Starting Grant COeXISTENCE\, whi
 ch bridges machine learning and transportation to simulate how intelligent
  machines compete with humans for limited urban resources\, such as space\
 , and what outcomes can be expected. He is also involved in the Horizon Eu
 rope project\, SUM\, and has served as Principal Investigator for the NCN 
 Opus Grant on Shared Mobility during Pandemic Times (2021–2024). Previou
 sly\, from 2019 to 2021\, he worked with Prof. Oded Cats at TU Delft on th
 e ERC Starting Grant Critical MaaS\, modeling two-sided mobility platforms
  with a specific focus on ride-pooling (ExMAS) and developing an agent-bas
 ed simulator for Uber-like systems (MaaSSim). Dr. Kucharski completed his 
 PhD at the Cracow University of Technology under the guidance of Prof. And
 rzej Szarata\, collaborating closely with Prof. Guido Gentile from La Sapi
 enza University of Rome on non-equilibrium dynamic traffic assignment.\nCo
 nnected and autonomous vehicles may fundamentally change how routing decis
 ions are made in urban networks. Today\, millions of human drivers indepen
 dently choose their routes\, learn from experience\, and collectively prod
 uce familiar congestion patterns and traffic equilibria. In the future\, s
 ome of these decisions may be delegated to machines capable of communicati
 ng\, learning\, and coordinating their actions across entire fleets.\n\nTh
 is seminar considers what happens when individual human route choice is gr
 adually replaced by collective\, algorithmic routing. Autonomous fleets ma
 y optimise their own travel times\, system performance\, market share\, or
  other objectives\, while interacting with human drivers and competing fle
 ets. This creates new routing games\, adaptation dynamics\, and notions of
  equilibrium that go beyond the classical traffic assignment paradigm.\n\n
 I will present recent results from the COeXISTENCE project (https://www.ra
 falkucharskilab.pl/research/coexistence/) illustrating both opportunities 
 and risks of this transition. We find that fleet strategies can substantia
 lly alter network performance and the distribution of benefits between hum
 ans and autonomous vehicles\; coordinated vehicles may form advantageous r
 outing coalitions\; and learning routing strategies with current multi-age
 nt reinforcement learning methods remains surprisingly difficult. At the s
 ame time\, socially aware routing and new forms of coordinated assignment 
 indicate that efficient and equitable outcomes may be attainable.\n 
LOCATION:GC B1 10 https://plan.epfl.ch/?room==GC%20B1%2010
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR