Driver ETA Estimation Using Cell-Based Supply Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating a driver's time of arrival (ETA) are computationally costly due to the need for real-time driver location updates, K-nearest neighbor searches, and spatial indexes, which consume high storage and often underestimate arrival times, leading to passenger cancellations.
Innovation Solution
A method and system that divide a search area into cells, counting the number of drivers in each cell to estimate ETA based on relative distances, using stream aggregation and novel spatial indexes to reduce computational and storage costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If K-nearest neighbor search and spatial indexes are used to find nearby drivers, then driver location accuracy is improved, but computational cost and storage requirements increase significantly
Solution Approach 1:
The system divides the service area into a grid of cells, where each cell stores aggregated driver count information. This segmentation transforms the complex problem of finding individual driver locations into a simpler problem of identifying which cells contain drivers, reducing computational complexity from O(n log n) to O(1) for cell lookup.
Solution Approach 2:
Instead of storing and searching actual driver location data points, the system creates a simplified copy represented by cell grid information containing only driver counts. This copy enables ETA estimation without requiring access to the original complex spatial database or performing K-nearest neighbor searches.
2Loss of information
If detailed spatial indexes are maintained for real-time driver tracking, then driver supply visibility is improved, but storage consumption increases
Solution Approach 1:
The continuous spatial space is segmented into discrete cells, transforming the storage requirement from storing individual driver coordinates to storing aggregated counts per cell. This reduces storage from O(n) for n drivers to O(c) for c cells, where c is typically much smaller than n.
Solution Approach 2:
The system changes the stored parameter from precise driver location coordinates to aggregated driver counts within cells. This parameter transformation maintains sufficient information for ETA estimation while dramatically reducing storage requirements.
3Measurement precision
If point-to-point ETA calculation is performed for each driver, then arrival time accuracy is improved, but processing time increases
Solution Approach 1:
The system pre-calculates and stores ETA values for each cell to the destination, rather than calculating point-to-point ETAs during runtime. This preliminary action allows the system to retrieve pre-computed cell ETAs and combine them with driver counts to estimate overall arrival time, avoiding repeated map engine queries.
Solution Approach 2:
The system merges multiple individual driver ETA calculations into a single cell-based ETA estimation. By aggregating driver information at the cell level and using cell-to-destination ETAs, the system combines multiple potential driver routes into one representative ETA value, significantly reducing processing time.
Data Source
AI summary
The present disclosure provides methods and systems for estimating a time of arrival of a driver at a location. In some examples, there is provided a method comprising: defining, by a server, a search area divided into a plurality of cells, the search area having a location as a centre point of the search area, the location being indicated in a request for a driver from a requestor device, the requestor device being a computing device associated with a user; determining, by the server, a number of provider devices located in each cell of the plurality of cells based on a location of each provider device of a plurality of provider devices, each provider device being a computing device associated with a driver located within the search area; and estimating, by the server in real time and in response to the request for a driver, a time of arrival of a driver at the location based on a relative distance between the location and the plurality of cells based on the determined number of provider devices.