JusticeTech System

With ETA system the system asks:

"If I assign this driver to this rider, how long will pickup take?"

Conceptually:

Driver location + Rider location + Road network + Traffic + Route restrictions ↓ Predicted pickup ETA

So: Driver A >> ETA 8 min Driver B >> ETA 3 min Driver C >> ETA 6 min Now the matching engine has much more useful information. This is one reason ride-hailing is fundamentally different from simply finding the nearest GPS coordinate.

Let's continue.......

You can use this concept to build your local Uber like application in the future

This is me analyzing a multi million dollars company and breaking down every system system 😄😄

Follow me....

So why did Uber needs a routing/ETA system and why ETA becomes extremely important?

Uber explicitly says its matching system considers real-world factors because the closest driver isn't always the quickest.

A simple distance algorithm chooses A But Uber wants something closer to B because B can actually reach you faster.

Suppose there are three drivers: Driver Straight-line distance Road ETA A 700 m 8 min B 1.2 km 3 min C 900 m 6 min

Let me proceed....

Is anyone still with me?

As a developer you have to be very very vast as the way you think Always think out of the box

But "nearby" doesn't mean "best" This is where things become really interesting.

A driver belongs to a particular cell based on their GPS position. So instead of asking "Which of 500,000 drivers is close to this rider?" the system can effectively ask "Which available drivers are in this geographic area and its neighboring areas?" That dramatically reduces the search space.

Think of the entire city like this: ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ ⬡ Each hexagon represents a geographic area.

If there are 500,000 drivers and thousands of requests arrive simultaneously, you'd perform enormous numbers of distance calculations. Instead, Uber uses geospatial indexing. One of Uber's major technologies for geographic indexing is H3, an open-source hierarchical hexagonal grid system. Uber has described using H3 to support marketplace analysis and optimization, including pricing and dispatch.

Imagine Lagos has 500,000 online drivers and you request a ride from Lekki Phase 1. A naive programmer might write: for driver in all_drivers: calculate_distance(driver, rider) That above lines of code simply means go through every driver I have, and calculate how far that driver is from the rider. That is terrible at Uber's scale.

How does this work? Follow me

Here is the biggest trick Uber implemented a logic which is "don't search every driver"

So when you press Confirm, Uber isn't starting from zero. It's querying a live representation of the driver marketplace.