Demand-Based Taxi Station Allocation via Grid Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optimal allocation of taxi stations and standby taxis across a city region is challenging due to the difficulty in collecting real-time demand and traffic information, making it hard to define an optimization objective and resulting in computationally expensive methods like linear programming for large problem sizes.
Innovation Solution
A framework that extracts trip data from vehicle operation data to estimate vehicle demands and travel times between tiles, using a modified fast clustering procedure to determine optimal station locations and the number of standby vehicles, minimizing total travel time and enabling timely on-demand transportation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional linear programming methods are used for optimal allocation, then allocation accuracy is improved, but computational cost increases significantly
Solution Approach 1:
The patent divides the continuous city region into discrete grid cells, transforming the complex continuous optimization problem into a discrete allocation problem. This segmentation allows the use of more efficient discrete optimization algorithms while maintaining allocation accuracy, thereby reducing computational cost compared to traditional linear programming approaches on continuous domains.
Solution Approach 2:
The patent transforms the optimization problem by changing parameters from continuous coordinates to discrete grid cell indices, and from minimizing travel time directly to maximizing a scoring function based on demand density and accessibility. This parameter transformation enables the use of efficient discrete algorithms while preserving the essential optimization objectives.
2Area of stationary object
If the entire city region is considered for allocation, then coverage is improved, but information collection difficulty increases
Solution Approach 1:
By dividing the large city region into many small grid cells, the system can collect and process information locally at each cell level rather than attempting to gather all information simultaneously across the entire city. This segmentation makes information collection more manageable and scalable while maintaining comprehensive coverage.
Solution Approach 2:
The patent employs a data-driven approach where the system automatically collects and processes trip data from existing transportation records without requiring manual surveys or additional infrastructure. The grid-based framework enables automatic aggregation of demand information from trip origins and destinations, reducing the difficulty of information collection while maintaining city-wide coverage.
3Loss of time
If more taxi stations are allocated, then response time is improved, but operational cost increases
Solution Approach 1:
The patent applies local quality by allocating taxi stations with different priorities and densities based on local demand characteristics. High-demand areas receive more stations while low-demand areas receive fewer or none, optimizing the overall system performance without uniformly increasing the number of stations city-wide. This localized approach improves response time where needed while controlling overall operational costs.
Solution Approach 2:
The patent implements a scoring-based allocation that identifies and prioritizes only the most critical grid cells for station placement. Rather than distributing stations uniformly or allocating to all potential locations, the method selectively places stations in areas that provide the greatest marginal improvement in response time, achieving satisfactory overall performance with fewer total stations.
Data Source
AI summary
Described herein is a framework for optimal demand-based allocation. In accordance with one aspect of the framework, trip data is extracted from vehicle operation data received from one or more vehicle data devices servicing a region of interest. Vehicle demands for multiple tiles of the region of interest and mutual travel times between neighboring tiles are estimated based at least in part on the trip data. A set of candidate tiles for allocating vehicle stations may then be determined by minimizing total travel time based on the mutual travel times and vehicle demands.