Methodological advancements in urban mobility systems

This talk presented new results on choice-driven matching for large-scale mobility platforms, addressing a problem that has become increasingly relevant as ride-hailing and micromobility services scale to city-wide demand: how to match riders to vehicles or routes in a way that respects individual preferences without making the underlying optimization intractable.

The proposed approach decomposes the matching problem into smaller subproblems that can be solved near-independently while provably converging to a solution close to the full-system optimum. Early experiments on simulated urban networks show substantial speed-ups over monolithic formulations with only a small loss in solution quality — a trade-off that matters in practice, since many mobility platforms need to re-match every few seconds.

The talk closed with open questions on how these methods generalize to mixed fleets of human-driven and autonomous vehicles.