Collaborative Mutil-agent planning of autonomous taxis: A Case Study in New York Manhattan island

Hello, everyone, welcome to the "New York autonomous taxi intelligent planning" projects.

Data source

The data used in this research were collected from taxi data-sets in Manhattan island, New York from November 1st-7st, 2019 on Yellow Cab's website https://www1.nyc.gov/site/tlc/about/tlc-trip-record-data.page.

We process the trip-related dataset in Data processing procedure, other data was avaiable on my Google Drive.

If you want to recur our work, you need to download the whole data to replace the ‘Data.file’ under this directory.

Simulation basic background

In the simulation phase,

• 3000 autonomous taxis are randomly generated in the network on 00:00 am per day.

• each day will be treated as an episode with 144 steps in the simulator.

• the study area is divided into 67 locations and comprised of 164 edges.

Benchmark heuristics

There are two baseline heuristics compared with our approach:

• Hotspot walk, essentially defines a stochastic policy for each agent, which indicates that the actions will be chosen according to the probability computed by trips quantity at each step.

• DiDi repositioning, employs a centralized optimization framework to determine the repositioning tasks(next pick-up location) of each taxi. Each task will be assigned by solving the Mixed-integer linear programming (MILP) model.