../multimodal-pathfinding

Multimodal Pathfinding

I was looking for jobs and a university was purposing a job about moultimodal pathfinding. Sound fun i need to dig google. Also i can’t sleep btw

[It’s just notes and it may change in the future]

Problem synthesis

I will try to find as much feature as possible some may be not realistic (yes i’m righting this after righting about weather optimised path)

You have a map of different scale, for better visualisation i will limitate it to the scale of a city.

We will take Amsterdam because it have various way of transportation.

Our goal is to see how we can optimise the traval of someone through different transportation system.

Score

Optimizing doesn’t mean anything alone, we need to define parameter to optimise, in this problem it can be :

Distance

Really easy to determine we just add the length of the different lines compostion the path and we get te length, even with the altitude variation it’s just basic math.

Price

For this you need the data from the different transportation services and the personnal data of the user, with that we will have our first privacy problematic, i would prefere that this data stay local.

Duration

If you are not google getting a good duration is very tricky because they have spyware in all our pocket to finetuned there result. I’m curious if it’s possible to do approximation without tracking user and just having his start and end time value. (may try simulation to check that)

Effort

Easy to calculate with public transport -> 0. But it’s a little subjective, however we can try to do it as good as possible with just minimising height variation and crowded places.

CO2

Hard to get precise data but to be good enough value we just need general statistics applied to the locallity.

Accessibility

Very hard to implement because of regular street works but with community entry it’s not impossible to have a good one.

Transportation mode

Generic

Walking, cycling, driving a car is of course possible in every city

Public

Here we are talking about Bus,Subway, Tram or Train.

For the Tram we may have some problem on CO2 calculation because some are hybrid and use different energy depending on the road portion they are.

Private

Taxi, Uber and alternatives of course. Uberlike are embed in maps so may be possible to implement it in an alternative.

There is also light vehicle rent like V’lib system or scooter. for the problem we will split it juste in 2 category park enywhere and park everywhere.

Fancy

In our exemple of Amsterdam you can go to the north with subway, train or ferry !

It’s important small optimisation to keep in mind to optimise the travel.

Data

The map

To get the data hopefully we have openstreetmap which have a gigantic crowdsourced database. Also they have some cool information to give to the user like the nearest water point toilet or shadow area. So it’s event possible to adapt the path to be as pleasant as possible according to the weather.

The transportation

Well of course for each city you can walk, bike and drive. But the public and renting transportation must be very accessible, if not nobody would use it, but the different price calculating may be harder for some renting system.

Crowded places

That the tricky data to get if we want to respect the privacy of the user. But there may be way like the system use during the covid to trace if you were in contact. If the app is largely spread you may be able to transmit through short range info about if place near are crowded or not, to the cost a bettery live. And the algorithme adapt in real time.

Algorithm

I know 2 pathfinging algorithm from memory, dijkstra and A*.

Intuition

Build a map with constant “weight teleportation” for public transport and running an A* through it to optimise the X parameter sound like a good intuitive option. BUT.

SotA

[will take time FUCK it’s complex]

Simulation

I think that to test theories it’s good to have a small simulation system to run algorithm in half realistic case.

Need :