Vector-based pedestrian navigation in cities

A fascinating paper that analyses how humans navigate cities on foot, and how their routes compare to the optimal ones.

It turns out that we’re remarkably good at finding near-optimal routes without resorting to complex calculations; we use heuristics and we “satisfice” for a good-enough route. But it turns out that, when our route planning is sub-optimal, it’s sub-optimal in predictable ways.

One such puzzling way we consistently differ from the optimal route is our tendency to pick asymmetrical routes; we walk a different route on our way to a destination than the one we walk coming back. This reveals a heuristic that we use, which the researchers called the “initial straightest segment” heuristic: we’ll set out on our journey by heading in a direction that’s as close to the direction of our ultimate destination as possible, even if the shortest route actually means taking a slightly different direction first. On the way back, we do the same thing – which often means picking a slightly different route.

A really neat example of using mobile data to discover fundamental aspects of human behaviour (and in a non-creepy way!)

Christian Bongiorno, Yulun Zhou, et. al. “Vector-based pedestrian navigation in cities”. Nature Computational Science 1(678–685), October 2021